Behavioural and metabolic responses of Unionida mussels to stress

Funding information Natural Environment Research Council; Scottish Alliance for Geoscience, Environment and Society Abstract 1. The aim of this study was to assess the extent to which the behavioural traits of freshwater mussels provide suitable indicators of stress in individuals, towards the advancement of non-invasive, remote monitoring techniques to examine population condition. 2. Variation in the expression of particular behavioural metrics was examined in accordance with measurements of oxygen consumption, across environmental stressors (aerial exposure and high concentrations of total suspended solids), and between two freshwater mussel species (Margaritifera margaritifera and Anodonta anatina) 3. Aerobic metabolic rate was quantified using intermittent respirometry, and behaviour was observed using time-lapse footage. Comparisons of metabolic response and the occurrence of behavioural traits, across the two stressors, focused on differences between the 24 h pre-exposure period (pre-exposure), the first 3 h of post-exposure (immediate post-exposure), and the time following the initial 3 h of post-exposure until the end of the experimental run (extended postexposure). 4. The results of this study demonstrated a relationship between the frequency of occurrence of behavioural responses to stress exposure, associated with valve activity, and significant changes in the metabolic functioning of A. anatina and M. margaritifera mussels. The findings from the study also highlighted substantial intraspecific variation across species and stressors. 5. Data from this research could assist in the development of novel biosensors that track mussel valve activity remotely in their natural environment. When coupled with real-time data examining alterations in environmental metrics, this technology could assist in the monitoring of population condition and aid conservation management.

One area of study that has witnessed increasing interest as a means of addressing these gaps in knowledge concerns the use of biomonitoring tools, or the tracking of specific biological processes and how these processes respond to alterations in the environment (Galloway & Depledge, 2001;Gagné et al., 2002;Blaise & Gagné, 2009;Farcy et al., 2013;Fritts et al., 2015). The study of biological responses may assist in detecting early warning signs before the occurrence of mortality (Handy & Depledge, 1999), provide a method to study the effects of sublethal stressors (Hartmann et al., 2016), and aid the evaluation of population condition in response to translocation and restoration efforts (Gray & Kreeger, 2014;Roznere et al., 2017;Salerno et al., 2018).
Behaviours reflect the response of an individual to a combination of environmental and physiological factors, and therefore have the capacity to provide sensitive, non-invasive indicators of stress in individuals (Robson et al., 2009;Hasenbein et al., 2015;Hartmann et al., 2016). Examples within the literature of non-lethal techniques for examining stress in freshwater and marine mussels often focus on two behavioural responses: movement (specifically, how a mussel may use its foot to move along the river bed or to burrow into the substrate- Johnson & Brown, 2000;Bartsch et al., 2010;Block, Gerald & Levine, 2013;French & Ackerman, 2014;Clements, 2015) and filtration (the active movement of water through the mantle, which facilitates respiratory and reproductive processes); previous studies suggest that both valve activity and clearance rates mirror individual responses to environmental change (Wilson, Reuter & Wahl, 2005;Nagai et al., 2006;Robson et al., 2012;Tuttle-Raycraft, Morris & Ackerman, 2017;Salerno et al., 2018). Both behaviours have the potential to provide an easy and cost-effective biomarker of stress (Kádár et al., 2001;Newton & Cope, 2006;Liao et al., 2009;Robson et al., 2009;Hartmann et al., 2016;Lummer, Auerswald & Geist, 2016), which could be scaled up to populations and species.
Despite interest regarding the use of behavioural traits as potential non-invasive indicators of stress in freshwater mussels, few studies have researched the physiological mechanisms that may drive their expression during stress exposure (Farcy et al., 2013;Archambault, Cope & Kwak, 2014). Research examining oxygen consumption rates ( _ MO 2 -a measure of aerobic metabolic rate) in other aquatic species has received significant attention as a method for testing hypotheses that relate variation in physiological traits with intraspecific variation in behaviour and life history traits (Biro & Stamps, 2010;Burton et al., 2011;Rosewarne, Wilson & Svendsen, 2016), with recent research demonstrating the efficacy of these techniques to evaluate stress in unionid mussels (Gibson, 2019;Haney, Abdelrahman & Stoeckel, 2020). In the study reported here, stress responses in freshwater mussels were examined, through the analysis of their behavioural traits, in conjunction with aerobic metabolic rate.
Very few studies that link the expression of behavioural traits with physiological condition in freshwater mussel responses to stress have examined individual variability (Hartmann et al., 2016). During stress exposure, animals may prioritize specific physiological functions and behaviours (Killen et al., 2013); however, the expression of a trait may not be consistent across species and individuals (Dingemanse et al., 2009;Biro & Stamps, 2010;Burton et al., 2011;Jolles et al., 2017). Specifically, the extent to which an individual prioritizes the expression of particular behavioural and physiological traits during stress exposure is thought to vary between conspecifics (Dingemanse et al., 2009). Therefore, to assess the consistency of metabolic and behavioural responses to stress across unionid species, the responses of two unionid species were examined: the freshwater pearl mussel, Margaritifera margaritifera; and the duck mussel, Anodonta anatina.
These species were chosen because they display similar distributional patterns and life-history traits yet appear to differ in their habitat requirements, which may assist in distinguishing species-specific thresholds in response.
Both M. margaritifera and A. anatina inhabit freshwater systems across Europeranging from the Iberian Peninsula in the south west, to Scandinavia in the north, and to Russia in the east (Zettler et al., 2006;Boon et al., 2019) and have been known to exist in sympatry, using the same host, brown trout (Salmo trutta;  (Bauer, 1988 Despite this, declines across M. margaritifera populations persist (Geist, 2010;Cosgrove et al., 2016;Lopes-Lima et al., 2016).
Recent work by Boon et al. (2019) (Cook, Wells & Herbert, 2011;Killen et al., 2013). To provide context to the expression of certain traits, individual responses should be observed across several environmental parameters, representative of common stressors a population experiences in the natural habitat, and presented at a magnitude necessary to evoke a response, thus determining whether the response is linear or has a threshold effect.
Consequently, this study aimed to compare the response of the two species across two stressors.
The purpose of this study was to investigate mussel behaviour as a biomarker for stress in unionid mussel species towards the creation of new techniques to assist in their conservation. To do so, this study tested the following three hypotheses: (i) the physiology of mussels, measured as oxygen consumption, shows a quantitative response to stressors; (ii) the expression of certain behavioural traits, measured as frequency of occurrence, shows a quantitative response to stressors; and (iii) behaviour can be used as a non-invasive, nondestructive biomarker of underlying physiology in freshwater mussel species.

| Experimental set-up
The experimental set-up ( Figure 1)

| Experimental overview
The experiment was designed to compare the physiological response (metabolic rate) and behavioural responses of the same individuals of two different mussel species to two different stressors: air exposure and suspended sediment ( Figure 1). A total of 40 mussels (20 from each species) were randomly selected for the experiment. Each individual was exposed to the two stressors or control conditions. There were four treatments for each of the emersion and TSS experiments, including one control condition F I G U R E 1 (a) Experimental set-up within each of the four holding tanks; (b) the conditions a mussel experienced during pre-exposure and two stressor exposures: emersion and increasing concentrations of total suspended solids (TSS) T A B L E 2 An overview of the stressors used to elicit a physiological and behavioural response in Anodonta anatina and Margaritifera margaritifera mussels Note: For the two stressors, the method for eliciting stress exposure and the variation in extent of stress exposure between the stressor magnitude groups is shown. For experiments examining heightened concentrations of total suspended solids (TSS) as a stressor, 'Polysperse 10' kaolin was added to the water of the holding tanks to achieve a desired nephelometric turbidity unit (NTU).
(see Table 2 for details of magnitude). Before, during, and after exposure to each stressor condition, the behaviour of each individual was quantified, with oxygen consumption rates recorded before and after exposure.
Each trial was conducted on four mussels simultaneously, each experiencing one of four treatment conditions (low, medium, and high stress magnitude and a control group), and consisted of six sequential steps: (i) a 2 day acclimation period; (ii) a 2 h background check; (iii) a 24 h pre-exposure period; (iv) a stress exposure (Table 2); (v) a minimum 18 h post-exposure period; and (vi) 2 h background check.
An individual experienced two trials, one with each of the two stressors. A period of 6 weeks rest was given to each mussel between the two trials, with marginal differences in individual standard metabolic rate (SMR) suggesting this was adequate for recovery (Table 3).
Before the experimental stress exposure commenced, all mussels were acclimated to an experimental temperature, 15 ± 0.5 C, for 2 days in a 30 L tank with untreated fresh water pumped from Loch Lomond and natural algal concentrations. Two background checks, undertaken in the absence of mussels, recorded oxygen reduction in the metabolic chambers for 2 h, before and after a trial, to obtain measures of microbial respiration (Svendsen, Bushnell & Steffensen, 2016). Oxygen concentration in the metabolic chambers during this period was regressed on time in both background check periods to quantify changes in background respiration over the course of a trial. Approximated background respiration was subsequently subtracted from measurements of mussel oxygen consumption. After the background check, individuals were placed in the corresponding metabolic chamber for a pre-exposure period. Here, mussels remained for 24 h undisturbed to record potential diurnal fluctuations in metabolic rate, and provide sufficient acclimation time (Gibson, 2019). Following this, mussels were exposed to the relevant stressor (Table 2). After stress exposure, mussels were left

| Oxygen consumption
Oxygen consumption ( _ MO 2 ; mg O 2 h À1 ) was measured using intermittent respirometry, using a computer-controlled set-up that recorded oxygen partial pressure and temperature (sampling rate, 10 s). Water oxygen content in the metabolic chambers was measured using optodes (FireSting 4-channel oxygen meters; Pyroscience, Aachen, Germany; www.pyro-science.com). Intermittent respirometry was conducted according to the technique described by

| Metabolic rate analysis
To determine an individual's metabolic rate as a proportion of metabolic tissue (mg O 2 h À1 kg À1 ) required metabolic tissue weights.
To obtain A. anatina metabolic tissue weights, mussels were euthanized at the end of the experiment. Harvested tissues were dried at 70 C for 2 days to provide final dry tissue weights.
Individual M. margaritifera were not euthanized to obtain dry tissue weights owing to their endangered status. Instead, wet tissue weights were estimated. Empty shells were collected from the sampled population. To estimate live shell weight, the relationship between shell length, width, and height and dry weight was calculated using a linear regression constructed from dead shells. Estimated live shell weight was then estimated from shell linear dimensions and subtracted from the total wet weight of live individuals, to estimate wet tissue weight for live M. margaritifera. Mass independent metabolic rates (MIMRs) were calculated to standardize metabolic rates and reduce the intraspecific variation (up to threefold differences in SMR were observed between conspecifics), using residuals from a regression analysis between SMR and tissue weight (P < 0.001) (Auer et al., 2015). Body mass and metabolic rates were log 10 -transformed before analyses to normalize and linearize the data.
Individual SMRs (mg O 2 h À1 kg À1 ) were calculated using oxygen consumption measures in the final 10 h of pre-exposure. Readings taken during this period, within 1 SD of the mean, were averaged to generate a final estimated SMR for the corresponding individual (see Table 3 for summary of calculations).

| Behavioural analysis
Behavioural analyses were conducted only on mussels experiencing high and medium stress magnitude. Underwater digital cameras

| Statistical analysis
Data were investigated using mixed effects models in R version 3.5.3 (R Core Team, 2020).

| Metabolic differential
Assessment of the metabolic differential revealed similarities between the two stressors and between M. margaritifera and A. anatina (Figure 3). A significant effect of stressor magnitude on the size of the metabolic differential during emersion experiments was shown in both species. For M. margaritifera, the differential was significantly higher in the low (P < 0.05) and high stressor magnitude (P < 0.05) groups than in the control group. Similarly, the metabolic differential was significantly higher in the low stressor magnitude groups than in the control (P < 0.05) in A. anatina. For TSS exposure experiments, no significant differences between stress magnitude groups and the control were found in M. margaritifera and A. anatina.

| Transition frequency
Examination of the transition frequency (alterations to a mussel's shell aperture width) as a behavioural response revealed some similarities

| Avoidance behaviour
Examination of avoidance (observations where the mussel's shell was closed) as a behavioural response showed some similarities at the interspecific level. There was no significant effect of stress magnitude The effect of stress magnitude on the extent of change in metabolic rate in response to stress exposure. A comparison of metabolic differential in Margaritifera margaritifera and Anodonta anatina following emersion and TSS exposure. The violin plot shows the mean and SE for the metabolic differential of corresponding individuals across the stress magnitude groups. The significant differences highlighted in the output of Kruska-Wallis tests, for comparisons of mean metabolic differential between stress magnitude groups and the control, are indicated by asterisks (*, P < 0.05;**, P < 0.01; ***, P < 0.001) during emersion and TSS experiments. Time had a significant effect on the occurrence of avoidance behaviour for M. margaritifera in emersion experiments, with avoidance behaviour significantly higher during stress exposure (P < 0.001) and post-exposure (P < 0.05) than during pre-exposure conditions. There was a significant effect of time for A. anatina in emersion experiments, but this was limited to a significant difference between stress exposure and pre-exposure

| DISCUSSION
The results of this study reaffirmed the notion that the behavioural response of unionid mussels to stress exposure provides a useful biomarker for examining the effects of environmental parameters on individual condition. Previous studies have established filtration and evasive behavioural strategies as biomonitoring tools to investigate tolerance to set concentrations of specific pollutants and between periods of rest and exposure (Tran et al., 2003;Liao et al., 2009;Hartmann et al., 2016;Haney, Abdelrahman & Stoeckel, 2020; Premalatha, Saravanan & Karuppannan, 2020). Nevertheless, this is The effect of stress magnitude on variability in metabolic rate of individual mussels over time. A comparison of metabolic variability (confidence interval (CI) of mean metabolic rate of individual during corresponding time period) in Margaritifera margaritifera and Anodonta anatina following emersion and TSS exposure. Each point corresponds to the mean CI (±SE) for the relative stressor, stress magnitude, species, and time. The significant differences highlighted in the output of the Kruskal-Wallis tests, for comparisons of the mean confidence intervals within stress magnitude groups and between the pre-exposure time period and the two post-exposure time periods, are indicated by asterisks (*, P < 0.05; **, P < 0.01; ***, P < 0.001) the first known study that has attempted to identify common behavioural responses in freshwater mussel species across multiple environmental stressors and to associate these with measures of physiological stress. The results of this study provide evidence of behavioural responses to stress exposure that can be linked to physiological condition, specifically to metabolic rate, in A. anatina and M. margaritifera mussels. The study also revealed substantial intraspecific variation, highlighting the importance of individual variability when examining stress response across populations.

| Metabolic response
A key component of this study was to examine whether physiology, measured as oxygen consumption, displayed a quantitative response to stressors. Initial findings showed significant differences in individual metabolic functioning, across both species, with mussels found to exhibit idiosyncratic metabolic responses to stress exposure: some mussels appeared to heighten their metabolic rates, whereas others displayed a metabolic depression immediately after stress exposure. However, significant individual variation was already present before exposure to the stressors, with a threefold and fourfold difference between the maximum and minimum values for SMR in A. anatina and M. margaritifera respectively, which is a common finding in many other aquatic species (Burton et al., 2011;Kristín & Gvoždík, 2012;Van Leeuwen, Rosenfeld & Richards, 2012;Metcalfe, Van Leeuwen & Killen, 2015). Therefore, quantification of physiological responses to stress required analysis that sought trends among the noise of individual variation.
Deviation from normal metabolic functioning for extended periods of time following stress exposure was found to be common across species and stressors, thus presenting metabolic variability as a potential method for quantifying response to stressors. This observed increase in metabolic variability and frequent failure to return to normal metabolic functioning within the experimental time limit following stress exposure is well documented in the literature: studies concerning metabolic response of bivalves to stress exposure provide evidence to suggest that individuals will sometimes require several days to return to pre-exposure levels (Newton & Cope, 2006;Robson et al., 2012;Ridgway et al., 2014;Lopes-Lima et al., 2016;Payton, Johnson & Jenny, 2016).
It is likely that both stressors used in this study would affect the physiological functioning of freshwater mussels. Emersion removes the appropriate medium for the mussel's specialized respiratory structures, consequently preventing filtration activity from fulfilling an individual's metabolic requirements. The subsequent establishment of an energy deficit may force a substantial reduction in energy dissipation to prevent fatal thermodynamic imbalance and cell death F I G U R E 5 The effect of stress magnitude on the occurrence of transition frequency over time. A comparison of the mean (±SE) proportion of time transition frequency occurring in Margaritifera margaritifera and Anodonta anatina following emersion and total suspended solids (TSS) exposure, across the medium-and high-stress magnitude groups. Significant differences (P < 0.05) highlighted in the output of Kruskal-Wallis tests (for comparisons of the mean proportion of time mussels display the behaviour) between the pre-exposure time period and the corresponding post-exposure time period are marked with α and β respectively (Widdows & Shick, 1985;Thomsen & Melzner, 2010). By contrast, high concentrations of inorganic suspended sediments are thought to increase the energetic demand of particle processing, with the active excretion of undesired compounds in pseudofaeces incurring an energetic cost to individuals (Vaughn, Nichols & Spooner, 2008;Lummer, Auerswald & Geist, 2016;Tuttle-Raycraft, 2018). For both stressors, the perceived deviation from the SMR following intense physiological activity is perhaps reflective of individuals' continued attempts to adjust their filtration rates to compensate for disturbance to osmoregulation (Hartmann et al., 2016), nutrient turnover (Lorenz & Pusch, 2013), and respiratory processes (Shick et al., 1986).
Previous studies, examining alterations to the clearance rates of freshwater mussels in response to TSS, have often alluded to a threshold in response at 8 mg L À1 , above which clearance rates are significantly diminished (Foster-Smith, 1975; Madon et al., 1998;Gascho Landis, Haag & Stoeckel, 2013;Tuttle-Raycraft, Morris & Ackerman, 2017). In this study, attempts to define a similar threshold in the physiological response of mussels to increased stress from prolonged emersion or heightened concentrations of TSS found no substantial differences between stress magnitude groups, only between mussels that experienced stress and those that did not. Alexander, Thorp & Fell (1994) discovered a similar response in metabolic rate to increasing TSS with Dreissenia polymorpha: acute exposure to suspended solids evoked a depressed metabolic rate; however, oxygen consumption did not cease or continue to decline at higher concentrations of TSS. Therefore, the results concerning physiological response to stress exposure suggest a binary response to the presence or absence of stress, contrary to a positive linear relationship between heightened response and greater levels of stress initially imagined.

| Behavioural response
In addition to examining individual physiology, this study also assessed whether behavioural responses to stress could be quantified.
For both species and stressors, transition frequency increased in occurrence in response to stress exposure. Exposure to terrestrial conditions and suspended fine particulate matter are likely to have constrained the capacity of mussels to function as filter feeders (Widdows & Shick, 1985;Shick et al., 1986;Alexander, Thorp & Fell, 1994;Tuttle-Raycraft, Morris & Ackerman, 2017). To endure emersion, the adoption of brief periods of air breathing may have assisted in the removal of metabolic by-products through aerial diffusion, such as anaerobically produced CO 2 , thus permitting the conservation of energy stores and consequently preventing early fatigue (Shick et al., 1986). To cope with increased suspended fine particulate matter during TSS experiments, a consistent alteration of valve activity would assist in modulating an individual's exposure to suspended solids. Exposure may incur damage to the filter-feeding apparatus with inorganic solids overloading the gut and gills, interfering with filter-feeding functions and efficient gaseous exchange (Alexander, Thorp & Fell, 1994). Despite providing a potential coping mechanism, brief periods of aerobic respiration during exposure to either stressor are unlikely to entirely relinquish the negative physiological effects of stress exposure: mussels may remain reliant on anaerobic pathways (De Zwaan & Wijsman, 1976), with periods of closure interspersing phases of aerobic respiration to prevent physiological damage (Liao et al., 2009).
The implementation of anaerobic pathways to compensate for the energetic requirements of an individual during stress exposure would necessitate a recovery period after the removal of the stressor, dependent on aerobic metabolism (Richards, Heigenhauser & Wood, 2002;Burton et al., 2011;Robson et al., 2012;Haney, Abdelrahman & Stoeckel, 2020). To assist recovery, a constant movement of the shell aperture may have acted to facilitate an augmented filtration rate, by pumping the water over the gills, thus providing a pathway for reducing the oxygen deficit incurred and F I G U R E 6 The average marginal predicted probability of transition frequency occurring during a 30 min 'closed state', in accordance with alterations in metabolic rate (MR). Results are shown for Margaritifera margaritifera and Anodonta anatina separately, owing to differences in the calculation of MR: calculations of MR in A. anatina used measurements of individual dry tissue weight, whereas calculation of MR in M. margaritifera used predicted wet tissue weights, confounding direct comparisons between species removing potentially harmful substances (Widdows & Shick, 1985;Robson et al., 2012). It would appear, therefore, that the increased occurrence of transition frequency during and after stress exposure reflects a propensity of mussels to use behavioural traits to cope with stressors; however, this application appears to be specific to the stressor, the species, and the metabolic scope of an individual.
Individuals that display heightened transition frequency in response to stress exposure may be more likely to recover more quickly and, therefore, display a prompt return to normal activities after stress exposure (Marras et al., 2010). However, for transition frequency to occur, an individual is required to generate frequent shell movement, which necessitates the use of adductor muscles and is therefore likely to be energetically demanding (Shick et al., 1986).
Individuals with a higher metabolic rate or aerobic scope are more likely to cope with the energetic requirements of transition frequency, and thus use this behavioural trait more often. Furthermore, individuals of the same species were collected from the same study site, suggesting that environmental conditions in the habitat were unlikely to shape the observable phenotypic variation in behavioural and physiological traits, provided that heritability of an individual's physiological profile is low (Burton et al., 2011).
Avoidance behaviour and foot extension were observed less frequently in mussels after stress exposure and varied between the two species, generating large zero-inflated data sets.

| Variation in species and stressors
The results from this study suggest species-specific responses to the stressors, often perceived to reflect differences in physiology (Gough, Gascho Landis & Stoeckel, 2012;Ganser, Newton & Haro, 2013;Haney, Abdelrahman & Stoeckel, 2020). A key driver of these differences could also be the environmental conditions the populations experienced in their natural habitats. The sample populations used in this study were collected from ecosystems displaying very different habitat characteristics. The lentic system from which A. anatina were collected was subjected to frequent water abstraction and displayed poor water quality, suggesting a potential tolerance to prolonged stress exposure and previous experience with both stressors. By contrast, M. margaritifera were taken from a mill lade, hydrologically connected to the main channel of the river, characterized by relatively consistent depth and flow conditions in addition to good water quality. This suggests there may have been differences in the sensitivity to the stressor (Hart, Miller & Randklev, 2019), with the M. margaritifera population less well adapted to the presence of the experimental stressors for extended durations (Lummer, Auerswald & Geist, 2016;Johnson et al., 2018).
Owing to the significant differences in habitat where the species samples were taken, and the potential for this to be a significant driver of individual responses, this study was limited in its propensity to tease apart species differences.
In addition to interspecific differences in response, this study also highlighted differences in response to the two stressors, perhaps reflective of differences in the magnitude of stress caused by each stressor. However, there were differences in how these stressors were induced for this study: mussels were handled during the emersion experiments but were not handled during the TSS exposure study. Handling mussels might have heightened the extent of stress individuals experienced during the emersion experiments. Without handling the control mussels it is difficult to quantify the extent to which this evoked stress within mussels. Nevertheless, evidence from the literature regarding the impact of short-term handling on individuals suggests this may be negligible (Miller, Rach & Cope, 1995;Gray & Kreeger, 2014;Ohlman & Pegg, 2020

| Implications for behaviour as a biomarker
The results from this study suggest that exposure to an environmental stressor can be detected by measuring transition frequency in unionid mussels. This study demonstrates a clear distinction in the presence of this behaviour between periods before stress exposure and following exposure, which can be linked to alterations in metabolic functioning. Measurements of transition frequency could, therefore, form the basis for a biomonitoring tool to detect the onset of stress in populations.
Recording the frequency of occurrence of this behaviour over time could assist practitioners in identifying when individuals are experiencing prolonged stress and requiring conservation intervention. This biomonitoring tool may also be deployed to aid relocation and restoration efforts towards the conservation of populations, with research concerning the use bioindicators of unionid fitness already having demonstrated the applicability of such approaches (Gray & Kreeger, 2014;Roznere et al., 2017). For example, studies acting as prerequisites to a translocation scheme could deploy a small subset of a population into a habitat of interest and subsequently conduct monitoring of transition frequency to assist practitioners in gauging habitat suitability.
To quantify the extent of stress caused, research must account for individual variability. To do so, laboratory experiments could identify the most responsive individuals within a sample of the population to act as indicators for overall population condition.
The thresholds for individual stress response could be identified by focusing on the presence of transition frequency across a variety of stressors in these indicator individuals, thereby accounting for population-specific variation in response.
This study focused exclusively on adult mussels, and hence did not account for variation in response across life stages. Research would suggest that juvenile mussels are perhaps more susceptible to environmental stressors such as heightened fine particulate matter (Geist & Auerswald, 2007;Geist, 2010; Tuttle-Raycraft, Morris & Ackerman, 2017), although, given their size, studies such as this may be difficult to replicate with a sample of juvenile mussels. Therefore, biomonitoring techniques reliant on the monitoring of transition frequency may be limited to adult mussels yet could be used as a proxy to infer juvenile population condition.
To identify the onset and frequency of occurrence of behavioural metrics, this study relied on direct observation using high-resolution camera technology. This provided a useful method of unrestricted categorization of behavioural traits but required extensive analysis of the image data, and this method would also be difficult to use in a field setting. The use of animal-attached remote-sensing technologies, such as Hall sensors, circumvents such issues and allows measurements of mussel valve movement (valvometry) to be acquired at high resolution and in real time (Nagai et al., 2006;Robson et al., 2012). Previous studies have provided evidence to suggest that both avoidance behaviour and transition frequency could be analysed using biosensor technology (Lorenz & Pusch, 2013;Hartmann et al., 2016;Lummer, Auerswald & Geist, 2016). However, this technology is currently limited to laboratory experiments and is yet to be tested in the field as a remote-sensing technique.

| CONCLUSION
To obtain information specific to certain species across variable, spatial, and temporal scales, in addition to being predictive, prescriptive, and scalable, ecologists must move away from 'long tail' scientific methods conducted by individual investigators over limited spatial and temporal scales and reliant on funding models that provide limited scope for collaboration (Heidorn, 2008;Hampton et al., 2013). The adoption of a context-driven approach to ecology, which examines the physical attributes of the ecological landscape, in addition to how the animals respond to changes in their habitat, is likely to provide appropriate data for enacting successful conservation management. Using remote sensing to detect the occurrence of transition frequency in indicator individuals may assist such an approach. Data to suggest how a population is responding to alterations in environmental conditions before, during, and after conservation management (e.g. river restoration and reintroduction schemes) could assist the quantification of project success, providing population-specific thresholds to identify when particular environmental variables begin to impair the condition of individuals. This article highlights the potential of this approach for contributing to the conservation of endangered freshwater unionid mussels.