‘Phototaxis’ in the absence of light? Locomotory patterns in unionid mussels

Light is a key environmental cue that drives behaviour and movement patterns of aquatic invertebrates (Bruce-white and Shardlow, 2011; Naylor, 1999; Tierney et al., 2017). Light stimuli can thereby determine the timing of movement activities as well as their direction, potentially triggering movements towards (positive phototaxis) or away from (negative phototaxis) the light (Enever et al., 2022; Kappes and Sinsch, 2005; Toomey et al., 2002; Uryu et al., 1996). Consequently, light patterns and availability strongly influence species' spatial distributions and the temporal dynamics of many aquatic invertebrate communities (Hobbs et al., 2021; Kessler et al., 2008; Lagergren et al., 2008).

In aquatic habitats, light can function as a ‘master’ cue that, through its availability (presence or absence), intensity and quality (i.e. wavelength), provides information on multiple potential drivers associated with aquatic invertebrate fitness and survival (Gaston et al., 2013; Zhang et al., 2011). Firstly, many predators of aquatic invertebrates rely on optical prey detection (Brönmark, 1994; Janssen and Corcoran, 1993; Link and Edsall, 1996; Richmond et al., 2004). Hence, a shift to nocturnal activities or diurnal migrations within water bodies can effectively decrease the risk of predation (Hobbs et al., 2021; Kessler et al., 2008; Kohler and McPeek, 1989). Secondly, the attenuation of light with depth may inform spatial orientation and the organisms' proximity to the water surface. Such information is essential for fitness and survival as, for example, it regulates exposure to flow velocities in rivers and influences the success of larval settlement (Gaston et al., 2013; Porter and Marsden, 2008; Thorson, 1964). Lastly, light intensity and quality are tightly associated with food availability and quality, which substantially influence the fitness of freshwater invertebrate consumers (Cointet et al., 2019; Guo et al., 2016).

In freshwaters, mussels of the Order Unionida represent the most diverse and most endangered bivalve taxon (Graf, 2013; Haag and Williams, 2014; Lopes-Lima et al., 2017, 2018; Williams et al., 1993). As ecosystem engineers, unionids provide habitat (e.g. mussel beds) to other invertebrates and play a crucial role in nutrient cycling and water purification (Hopper et al., 2021; Vaughn and Hakenkamp, 2001; Vaughn and Hoellein, 2018; Zimmerman and de Szalay, 2007). While unionids are often perceived as sessile or sedentary organisms, vertical and horizontal movements can be essential for the survival and fitness of this taxon. For example, vertical movements can mitigate against ectoparasite infestations (Nichols and Wilcox, 1997). Horizontal movements are well-recorded mitigation strategies against desiccation risk when water levels drop (Curley et al., 2022; Johnson et al., 2001; Lymbery et al., 2021) and have reproductive functions, such as aggregation for increased fertilisation success and migration to suitable locations for larval release (Amyot and Downing, 1998; Shelton, 1997; Vicentini, 2005).

The movements of unionids are influenced by factors such as temperature, seasonality, changes in hydrological conditions, food availability and water quality (Allen and Vaughn, 2009; Block et al., 2013; Bovbjerg, 1957; Curley et al., 2022; Johnson et al., 2001; Levine et al., 2014; Lymbery et al., 2021; Nichols and Wilcox, 1997; Reynolds and Guillaume, 1998; Zapitis et al., 2021a). However, we still have a limited understanding of what drives unionid movement behaviour (Haag, 2012). While movements in response to light cues have been recorded in a number of marine and freshwater bivalve species – such as in king scallops, Pecten maximus, zebra mussels, Dreissena polymorpha, and golden mussels, Limnoperna fortune – and encompass both positive and negative phototactic responses (Enever et al., 2022; Toomey et al., 2002; Uryu et al., 1996), the role of light as a potential stimulus of unionid locomotion is largely unresolved. Yet, light availability and intensity have been shown to affect non-locomotion unionid behaviours such as valve closure (Braun and Job, 1965; McIvor, 2004) and to trigger changes in feeding rates (Hills et al., 2020). Hence, as light can convey information pivotal for informing microhabitat selection associated with predation risks, reproductive success and food availability, it is crucial to understand the influence of light on unionid movement and microhabitat selection.

In this study, we examined the light dependency of horizontal movement patterns in unionid mussels using controlled laboratory experiments with Anodonta anatina and Unio pictorum, two of the most abundant unionids in Europe (Lopes-Lima et al., 2017). Specifically, we assessed (i) the impact of light availability on locomotory activity, (ii) the light-directedness of locomotion (i.e. the movement towards or away from the light source) and (iii) the influence of light intensity on movement distance. Furthermore, we recorded mussel opening and the impact of shell size on locomotory distance. Our experiments allowed us to further elucidate the factors driving unionid microhabitat selection and gain insights into their ecological implications.

We collected 30 Unio pictorum (Linnaeus 1758) specimens from Mapperley reservoir (Ilkeston, UK; January 2018) and 45 Anodonta anatina (Linnaeus 1758) specimens from Markeaton Brook (Derby, UK; July 2018). Each species was kept separately in dechlorinated tap water (carbonate hardness 4–6°KH, pH 7.8–8.2, calcium concentration 80–120 ppm) in a 120 l recirculating system with a 1200 l h⁻¹ power head and a 1800 l h⁻¹ pump connected to an external filter removing nutrients (see Supplementary Materials and Methods for details).

Water quality was monitored weekly. Temperature, pH, oxygen concentration and oxygen saturation were measured with a Hach HQ40d multimeter (IntelliCAL™ LDO101 and pH Liquid PHC30101 probes, Hach company, Loveland, CO, USA). Ammonium (NH₄⁺), nitrate (NO₃⁻), nitrite (NO₂⁻), phosphate (PO₄³⁻) and calcium (Ca²⁺) ion concentrations, as well as carbonate hardness (KH) were measured with API Fishcare Kits (Mars Incorporated, McLean, VA, USA). We exchanged 20–30% of the water volume weekly, and increased exchange rates when nutrient concentrations exceeded thresholds of NH₄⁺>0.2 mg l⁻¹, NO₃⁻>10 mg l⁻¹, NO₂⁻>0.0 mg l⁻¹ or PO₄³⁻>1.0 mg l⁻¹.

A 48 inch Arcadia Original Tropical Fluorescence T8 lamp (F036; Arcadia Aquatic, Ely, UK) with a luminous flux of 1110 lm provided light from 06:00 h to 22:00 h. We initially kept U. pictorum at 10.5±0.5°C and gradually increased the temperature to 19.5±0.5°C. Anodonta anatina were collected in July and kept directly at 19.5±0.5°C.

The main food source provided was the green-alga Chlorella vulgaris (CCAP 211/74), which was cultured in batch cultures with Jaworski's medium (https://www.ccap.ac.uk/index.php/media-recipes/) on a shaking table. Chlorella vulgaris concentrations were determined spectrophotometrically with a Cecil CE 1011 (Cecil Instruments Ltd, Cambridge, UK) based on a previously established absorption to ash-free dry mass (AFDM) calibration curve (Zapitis et al., 2021b). Food was supplemented weekly by natural seston collected with a 20 μm plankton net from Markeaton Brook. No mortality occurred between the collection of mussels and the end of experiments in July 2018.

We conducted two experiments; in both, we assessed horizontal mussel movements, hereafter referred to as locomotion, and disregarded vertical burrowing activities. In Experiment 1, we assessed the influence of light availability, defined as the presence or absence of light, on (i) the proportion of mussels that engaged in locomotion and (ii) the locomotory path distance in U. pictorum. Anodonta anatina were not available at the time (see Supplementary Materials and Methods, ‘Anodonta anatina availability’). Locomotory path distance refers to the total length of the movement trail created in the sediment by an individual’s locomotory activity. In Experiment 2, we assessed the influence of light direction as well as intensity on locomotion behaviour (locomotory proportion and path distance) in U. pictorum and A. anatina.

Experiments 1 and 2 were both conducted in conditions similar to those for animal husbandry, in dechlorinated tap water at 19±1.0°C with light provided from 06:00 h until 22:00 h by a 48 inch Arcadia Original Tropical Fluorescence T8 lamp (F036; Arcadia Aquatic). The substrate in the tanks consisted of 3–5 mm gravel of 8 cm depth. Based on our experience during husbandry conditions, this depth sufficed for mussels to burrow fully in the sediment. During the experiments, we used light intensity and temperature data loggers and monitored these parameters at 1 min intervals (Hobo Pendant UA-002-08, Onset Computer Corporation, Bourne, MA, USA). We measured luminous flux, in photometric units, as a proxy for full radiometric spectrum. Furthermore, water chemistry was measured following the methods outlined above and the data are presented in the Supplementary Materials and Methods (see ‘Water quality and light availability readings’). Air stones provided aeration for 15 min every hour to oxygenate the water and homogenise the algal distribution.

Chillers were used to maintain the water temperature; these can cause vibrations (DC750 Refrigerated Aquarium Chiller, D&D The Aquarium Solution, Ilford, UK), which are known to influence bivalve behaviours such as shell opening (Roberts et al., 2015). We accounted for such effects by placing the chillers on Styrofoam plates, which absorb vibrations. We also placed the aquaria on separate Styrofoam plates to provide an additional buffer. Additionally, we examined the potential influence of the chillers in preliminary experiments and did not find any effect on the direction of the movement.

The experiment examining the impact of light availability on mussel locomotion was replicated 9 times (n=9) in 75×43×40 cm tanks filled with 50 l of water. Experimental runs were conducted in May 2018 and each lasted for 47 h (starting at 07:00 h). Experiments were set up using a 50:50 mix of dechlorinated, temperature-adjusted water and water from the tank in which mussels were housed, to both maintain low nutrient concentrations and avoid transfer shock. We then placed eight randomly selected mussels in a vertical position with 30% of the shell covered by the substrate at an equal distance along the central axis of the tank directly under the light tube (Fig. 1A). At the start of the experiment, and then daily at 14:00 h, 22:00 h and 06:00 h we provided C. vulgaris as food at a concentration of 0.5 mg AFDM l⁻¹. The concentration was selected based on previous findings of higher locomotory activity at concentrations lower than 2 mg AFDM l⁻¹ in both A. anatina and U. pictorum (Zapitis et al., 2021a). Light was provided from above (Fig. 1A).

Mussel behaviour was recorded by vertical time-lapse photography at 1 min intervals (GoPro Hero5 black camera). Further, we recorded the position and new trails of mussels at all feeding times and at the end of each experiment. Temperature and light intensity loggers collected data throughout the experiments. Oxygen concentration and saturation, pH and carbonate hardness were recorded at the start and end of each experiment.

Experiment 2 was conducted in 200 l tanks (155×55×50 cm) in July 2018 to evaluate the impact of light intensity and direction on mussel locomotion. An experiment that lasted for 4 days was conducted for each mussel species. Light was provided laterally at a height of 20 cm (Fig. 1B). Light intensity was recorded along the length of the tank. Measurements were used to establish a light attenuation regression model that describes the decrease in light intensity with increased distance from the light source in the tank (see Fig. S1). We placed 24 conspecific mussels in three rows (8 per row) at three distinct distances from the light source and measured the distance using vertical photography. We then used the distance and the light attenuation model to calculate the light intensity at the start of the experiment for each specimen. The groups are referred to as high intensity (HI), intermediate intensity (II) and low intensity (LI) and their mean±s.d. light intensity was 1203±28 lx (distance of 29.0±1.0 cm), 450±20 lx (distance of 80.0±3.0 cm) and 228±8 lx (distance of 115.0±3.0 cm), respectively. Light intensity in the tank ranged between 142 lx (the end of the tank) and 1867 lx (near the light tube).

At the start of the experiment (13:00 h) and every 24 h, we fed mussels with C. vulgaris at a concentration of 0.5 mg AFDM l⁻¹. Before feeding mussels, we recorded the valve opening behaviour and categorised mussels with open valves and visible inhalant or exhalent siphons as ‘open’, and with no visible siphons as ‘closed’. We also measured the water temperature, oxygen concentration and saturation, pH, calcium ion concentration and carbonate hardness. We took vertical photographs and merged them into a single image using Adobe Photoshop CC 2018 (Adobe Inc., San Jose, CA, USA; see Fig. S1). From these images, we measured changes in the position of the umbo in relation to the light source, and the path distance covered. Horizontal time-lapse photography at 1 min intervals was used to record the time of activity (Fig. 1B).

We split each day of the 47 h experiment into a light period starting before noon (morning period M, 06:00 h to 14:00 h), a light period starting after noon (afternoon period A, 14:00 h to 22:00 h) and a dark period (night period N, 22:00 h and 06:00 h). An exception was the first time period on the first day of each experiment, which started at 07:00 h. For each period, we evaluated the proportion of mussels that engaged in locomotion. The influence of light availability (presence/absence) on the proportion of mussels that engaged in locomotion in a given time period was analysed with Pearson's χ²-test, accounting for data heteroscedasticity. The first morning period (M1) was removed from the analysis as handling of mussels during their placement in the aquaria can induce movement and bias the results (Uryu et al., 1996). The path distance for M1, which lasted 7 h, was standardised for the 1 h shorter duration compared with all other 8 h time periods by dividing each individual's path distance by 7 and multiplying it by 8. The influence of light availability on movement path distance was analysed using an ANOVA, with day of the experiment as nesting factor. Further, we analysed the influence of shell length on (i) path distance with a generalised linear model (GLM) with a quasi-Poisson error structure that was applied because of overdispersion, and (ii) the proportion of mussel movement per time period (out of six time periods in each experimental run) with a logistic GLM model.

We applied Pearson's χ²-test to assess differences in the number of locomotory activities between the light and dark periods, separately for each species. We then applied a MANOVA approach for circular data as described in Landler et al. (2022) to evaluate whether there was significant prevalence in the direction of movements, either towards or away from the light source. For this approach, the sine and cosine of the mussels' movement direction (in radians) were used as response variables in a MANOVA. This allowed us to also include species identity and the starting position of mussels as additional predictors and use an Akaike information criterion (AIC)-based approach to determine the best-fitting model structure (for updated resources concerning this approach, see https://github.com/Malkemperlab/Circular-MANOVA).

Further, we investigated potential factors that influenced the path distance of moving mussels. We considered light intensity and species identity as potential predictors and used a GLM model including a zero-inflation term to establish a linear regression model using the glmmTMB package (Brooks et al., 2017). The zero-inflation term was required (P<0.01), but zeros were also dependent on an interaction between light intensity and species identity (see Fig. S2). Unfortunately, the inclusion of these factors in the model's zero-inflation term resulted in model convergence issues and a zero occurrence structure could not appropriately be included in the model. As a consequence, the model selection processes become less reliable (Zuur et al., 2009) and in addition to AIC values we used a visual evaluation to determine the best-fitting model (see Fig. S2 for details).

All statistical analyses were conducted in R version 3.5.1 (http://www.R-project.org/). Regression models were assessed using standard residual diagnostics (Zuur et al., 2009). The most parsimonious model was selected based on AIC. Heteroscedasticity in linear models was accounted for by applying weighted generalised least squares (GLS) extensions.

In both experiments 1 and 2, the temperature and other water quality readings (i.e. pH, oxygen concentration and saturation, ammonium ion, carbonate hardness) fell within the anticipated limits. Furthermore, the light sensors confirmed that the intensity was consistent in the light period and no light was recorded in the dark periods (detailed values are given in Fig. S1).

The proportion of mussels that engaged in locomotion (locomotory proportion) was generally low (ranging between 0.0 and 0.5, i.e. 0–50% of the mussels moved; Fig. 2A), but was significantly higher in dark than in light periods (χ²-test, χ²₁=8.43, n=45, P=0.004, first time interval excluded because of handling bias; Fig. 2B). The median path distance was similar across all time intervals (range of medians: 20–24 cm) and light availability did not significantly influence path distance (nested ANOVA, F_1,12=0.013, P=0.909; Fig. 2D). Hence, once a movement event was initiated, the distance covered was independent of the presence or absence of light. Likewise, shell length, which ranged between 82 and 113 mm, showed no significant influence on locomotory proportion (logistic regression, P=0.093; Fig. 2E) and locomotory path distance (GLM, P=0.57; Fig. 2F).

In accordance with the findings of experiment 1, both species showed a significantly higher locomotory proportion in the absence of light (χ²-test: A. anatina: χ²₁=49.1 n=288, P<0.001; U. pictorum: χ²₁=11.7, n=216, P<0.001; Fig. 3A). Across the entire experiment, 100% of movement events by A. anatina and 78.6% by U. pictorum took place in the absence of light. Furthermore, over the 4 day experiment, both species moved towards the light source (Fig. 3B, Table 1). The direction of movement showed a significantly positive association to the direction of the light source in both species [MANOVA: circular mean±s.d. 185.8±52.5 deg (0 deg indicates the direction away from the light source), Pillai’s trace value=0.219, F=6.44, d.f.=46, P<0.01; Fig. 3C]. Neither species identity nor the light intensity group (i.e. high, medium or low light intensity exposure) affected the degree of movement towards the light source (model with lowest AIC did not include these factors).

Furthermore, despite the movement in the absence of light, light intensity during the light period significantly influenced the locomotory path distance in A. anatina, whose path distance decreased with increasing light intensity (Fig. 3D; Fig. S2). In contrast, no such relationship was found for U. pictorum (GLMM: A. anatina intercept 6.3, P<0.001; U. pictorum intercept 4.7, significantly lower, P=0.002; A. anatina species–light intensity interaction factor −0.0014, P=0.048; U. pictorum species–light intensity interaction factor −0.0004, P=0.33). Finally, we found no significant relationship between light intensity and the number of mussels that had their valves open. However, we recorded a decreasing number of mussels with open valves over time and a generally lower opening for U. pictorum compared with A. anatina (logistic regression: P<0.001; Fig. S3, Table S1).

Movements of unionids have been frequently recorded in the field (Amyot and Downing, 1998; Lymbery et al., 2021; Nichols and Wilcox, 1997; Saarinen and Taskinen, 2003; Schwalb and Pusch, 2007; Vicentini, 2005) but, beyond hydrological alterations, the drivers of locomotion are largely unresolved. In our experiments, we encountered similar path distances to those recorded for A. anatina and U. pictorum in the field (Saarinen and Taskinen, 2003; Schwalb and Pusch, 2007) and showed that the locomotion of unionid freshwater mussels is strongly shaped by light conditions. Both species primarily moved in the absence of light but showed a positive net movement towards the position of the light source, resulting in light-directed movements in the dark. In A. anatina, despite the mussels moving exclusively in the absence of light, their earlier exposure to higher light intensities during the light periods resulted in reduced locomotory path distances, suggesting that dim light conditions provide a stronger cue. Hence, our results suggest that light cues play an important role in mediating unionids' habitat selection, which can drive nutrient cycling and ecosystem structure in freshwater ecosystems (Hopper et al., 2021; Zimmerman and de Szalay, 2007).

Nocturnal locomotion frequently represents a predator avoidance strategy that reduces the predation vulnerability by visual foragers in many stream invertebrates (Flecker, 1992; Huhta et al., 2000; Oberrisser and Waringer, 2011; Ramírez and Pringle, 1998). We found that both A. anatina and U. pictorum had a significantly higher movement frequency at night and reduced movements during the day. As the extension of the foot during locomotion is associated with a higher predator vulnerability, the nocturnal movement may be an effective strategy to avoid predation by visual foragers, such as fish and diving birds feeding on unionids and other bivalves (Borcherding et al., 2013; Coughlan et al., 2017; Pietz and Buhl, 1999). Hence, predation may be an essential factor driving diel trends in unionid movements, highlighting the need for further research on the influence of predator cues on locomotory behaviour.

The primarily nocturnal locomotory activity we recorded contradicts previous findings of no substantial diel movement patterns in the unionid eastern elliptio, Elliptio complanata, in a lentic habitat (Amyot and Downing, 1998). One possible explanation for these diverging findings is a difference in the composition of the substrates. In contrast to our study, the locomotory behaviour of E. complanata was assessed for populations inhabiting soft substrates (shallow sandy beaches of low slopes; Amyot and Downing, 1998). In soft substrates, unionids can move horizontally in the sediment, remaining partially covered and protected by the substrate (Zapitis et al., 2021a). Our specimens originate from hard substrata: A. anatina from hard mixed-sized substrate and U. pictorum from either solely pebbly or compacted clay substrate. In hard substrata, mussels may crawl on the sediment and get exposed to higher predation risk during the day. This may explain the nocturnal activity we recorded. Interestingly, previous records on the populations examined in our experiments showed that A. anatina primarily crawls on the sediment while U. pictorum primarily moves in the sediment (Zapitis et al., 2021a). This may further explain the more profound preference for nocturnal locomotory activities recorded for A. anatina.

Another explanation for the diverging findings is the plastic response many populations of freshwater invertebrates display in environments with different predation risks. For example, the frequency of drift entries by mayfly larvae, a proxy of movement activities, is higher during daylight hours in predator-free environments (Ramírez and Pringle, 1998). However, in the presence of predators, activity patterns shift towards higher nocturnal drift frequencies, showing behavioural adaptations resulting in intra-specific variation (Flecker, 1992; Huhta et al., 2000; Oberrisser and Waringer, 2011). Whether bivalves also show such plastic population responses requires further investigation and predation-dedicated assessments.

Previous studies reported shoreward movements by A. anatina, U. pictorum and Unio crassus (Schwalb and Pusch, 2007; Zajac and Zając, 2011; Zając et al., 2016), and the escape of unionids from Potamogeton sp. beds in in situ conditions (Allen, 1923). However, the environmental drivers resulting in the above patterns could not be resolved under field conditions. Our findings under controlled laboratory conditions indicate that direction and intensity of light are important environmental cues regulating locomotory behaviour.

Taxis is commonly referred to as a direct response to an environmental stimulus, which in the case of positive phototaxis represents a light-directed movement (Riggs and Hoff, 2019). However, in the case of unionid mussels, the observed behaviour appears to be more complex as the response followed after the stimulation had ended. We refer to this delayed response as post-exposure phototaxis. Such behavioural patterns might be physiologically explained by two alternative hypotheses.

Firstly, the delayed response may be caused by a suppression of taxis in the presence of the light. An example of such a suppression of phototaxis is given by the silkworm larvae, Bombyx mori, where olfactory stimulation by mulberry leaves suppresses phototaxis (Inoko et al., 1981). Alternatively, the observed response may represent a learned mechanism, which would require the storage of information in a memory. Learning and memory has been already demonstrated in molluscs. For example, Lymnaea sp. freshwater snails can show ‘conditioned taste aversion’ and reduce their feeding activity when detecting non-detrimental sucrose previously paired and associated with detrimental salt (Sherff and Carew, 2009). Also, Aplysia sp. sea slugs show increased responsiveness following a harmful stimulus (sensitisation) and reduced responsiveness following repeated weak stimulation (habituation) to tactile stimuli (Sherff and Carew, 2009), and the marine nudibranch, Tritonia diomedea, exhibits learning behaviour in a locomotion context, demonstrated in its shoreward orientation (Brown, 1998; Willows, 1998). However, independent of the underlying mechanism (delayed taxis response or a learned behaviour), many field observations of locomotion are linked to light gradients (Allen, 1923; Schwalb and Pusch, 2007; Zajac and Zając, 2011; Zając et al., 2016) and hence light probably represents a key factor for microhabitat selection in unionids.

There are a number of potential factors that may underly light-directed movements and explain a positive post-exposure phototaxis in unionids. First, greater light availability is often transmitted in higher primary productivity, which can increase both food quantity and food quality (Singh and Singh, 2015; Guo et al., 2016; Cointet et al., 2019) and thereby mediate larvae production and bivalve reproductive success (Helm and Bourne, 2004). These relationships might, for example, explain the escape of unionids from macrophyte beds (Allen, 1923), which decrease light availability and may reduce the phytoplankton-food availability for filter feeders (Mulderij et al., 2007; van Donk and van de Bund, 2002).

As unionids feed on a range of seston particles, including microalgae, bacteria and fine particulate organic matter (Christian et al., 2004; Parker et al., 1998; Vuorio et al., 2007), the impact of light on food availability is likely to be less of a factor in streams with low phytoplankton densities and turbulent water flow. However, the original habitat of the A. anatina population used in our experiments is a low flow velocity brook that often becomes stagnant or even dries in the summer. Under such conditions, the coupling between light availability and primary production is again likely to affect unionid locomotory behaviour. Such interpretations are supported by earlier findings of lower movement likelihood at high food concentrations (Zapitis et al., 2021a), which motivated us to select realistically low food concentrations in our experiments.

Additionally, the reproductive success of unionids is also linked to planktivorous fish, which need to come into contact with glochidia larvae – for example, through ingestion – to ensure successful mussel reproduction (Haag, 2012; Modesto et al., 2018; Strayer, 2008). The prey detection distance of zooplanktivorous fish is well known to increase with light intensity (Link and Edsall, 1996; Modesto et al., 2018; Richmond et al., 2004). The light intensity in our experiments (between 100 and 2000 lx) aligns with the range over which fish predation reactive distance increases (Link and Edsall, 1996). Hence, the light-directed movement of unionids may also enhance the attachment rates of their glochidia on the gills of fish. Nonetheless, in Britain, neither A. anatina nor U. pictorum are known to release their glochidia in July (Aldridge, 1999; Lopes-Lima et al., 2017), and this hypothesis needs further testing during glochidial release times.

Therefore, mussel movement might be linked to a trade-off between reproduction and survival. Movement towards light is likely to increase reproductive success but might at the same time increase predation risk, which can be mitigated by nocturnal locomotion and/or by burrowing in the sediment during the day. Furthermore, the dependency of movement path distance on light intensity that we observed in A. anatina indicates that the benefits associated with light-directed movement level off at higher light intensities.

However, under certain conditions also, ‘win-win’ situations for mussels might emerge. Such situations might be encountered when inter-guild predation by visual top-predators on intermediate predators has a net-positive effect on survival (Kessler et al., 2008; Lagergren et al., 2008). For example, fishes and birds that prey directly on bivalves (Borcherding et al., 2013; Coughlan et al., 2017; Pietz and Buhl, 1999) also prey on many crayfishes, which in turn frequently prey on unionids (Durán-Lizarraga et al., 2001; Meira et al., 2019; Zając, 2014). Consequently, many crayfish avoid areas with higher light intensities to reduce their own predation risk, which is reflected, for example, in the light-induced increase of oxidative stress in red swamp crayfish, Procambarus clarkii (Durán-Lizarraga et al., 2001). Hence, under such conditions, increased light availability might not only be beneficial for reproduction but also reduce the overall predation risks for unionid mussels.

Finally, phototactic responses seem to be taxonomically widespread in freshwater bivalves, but the nature of the responses varies between taxa. For example, contrary to our findings on unionids, other families such as the Dreissenidae and Mytilidae often show a negative phototactic response (Toomey et al., 2002; Uryu et al., 1996). These opposing responses may partly emerge from different settling strategies of larval stages. The ovoviviparous Sphaeriidae give birth to their fully developed offspring in their own habitat (Heard, 1977; Mackie, 1978). Mytilidae and Dreissenidae can actively select their microhabitat as veligers (Porter and Marsden, 2008). However, unionids drop as juveniles from the gills of fish without selecting their microhabitat and, as juveniles, they feed by pedal feeding in the sediments (Haag, 2012; Strayer, 2008). Once their filter feeding phase starts, unionids may need to select their microhabitat and this positive phototactic response may enhance their fitness. Furthermore, the delay in the phototactic response, which is not recorded in other freshwater bivalve taxa, may be associated with their size and survival. Many unionid species are considerably larger than Sphaeriidae, Mytilidae and Dreissenidae (Killeen et al., 2004). This may make their movements more conspicuous and detectable by visual predators, resulting in the necessity of delayed phototactic responses.

Markeaton Brook and the shore of Mapperley reservoir are covered by trees. The light intensity range in experiment 2 reflects intensities at the surface of the water under vegetation: high intensity (∼1200 lx) represents midday summer conditions, intermediate intensity (∼450 lx) midday winter conditions, and low intensity (∼230 lx) represents late afternoon conditions (summer or winter; Bhandary et al., 2021).

Although in experiment 2 we provided light from a single side of the tank, we do not expect non-phototactic, e.g. magnetotactic, drivers to have affected the direction of movement. Also, we previously kept mussels in this system (with the light placed on the top) and we did not observe movements towards any particular direction. Nonetheless, experiments with the light provided from different directions would further validate our findings. Finally, we measured luminous flux as a proxy for the full radiometric spectrum. As invertebrates sense beyond the visible spectrum (Wasserman, 1973), further experiments could investigate light directedness in response to radiance and different sub-spectrums.

Our findings demonstrate the importance of light availability and intensity on unionid locomotory behaviour. The primarily nocturnal activity combined with the movement towards the light source in periods of no light availability indicates potential mussel memory and predator avoidance behaviour. We hypothesise that light functions as a sort of ‘master’ cue transmitting the information of multiple ultimate drivers of unionid microhabitat selection, and microhabitat selection is subject to a trade-off between the enhanced fitness associated with increased productivity and food availability and survival associated with the predation risk by visual predators. As incident light direction changes seasonally, it may affect unionid distribution and their impact on sediment mixing. Hence, the importance of light on unionid distribution in comparison to other environmental factors requires further investigation.

We conducted the experiments at the Aquatic Research Facility at the University of Derby, UK. This research formed part of the PhD work of C.Z. (Zapitis, 2020). We wish to thank Angie E. Ward for assisting with animal husbandry, Matt Gilooly and David Bryson for providing the cameras and technical support on time-lapse photography, and Petra Parmová for assisting with data collection.