Growth and longevity of the endangered freshwater pearl mussel (Margaritifera margaritifera) : Implications for conservation and management

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY-NC 4.0 https://creativecommons.org/licenses/by-nc/4.0/ Growth and longevity of the endangered freshwater pearl mussel (Margaritifera margaritifera) : Implications for conservation and management © 2024 the Authors Published version Nykänen, Sabrina; Taskinen, Jouni; Hajisafarali, Mahsa; Kuparinen, Anna Nykänen, S., Taskinen, J., Hajisafarali, M., & Kuparinen, A. (2024). Growth and longevity of the endangered freshwater pearl mussel (Margaritifera margaritifera) : Implications for conservation and management. Aquatic Conservation : Marine and Freshwater Ecosystems, 34(6), Article e4205. https://doi.org/10.1002/aqc.4205 2024 ARTICLE Growth and longevity of the endangered freshwater pearl mussel (Margaritifera margaritifera): Implications for conservation and management Sabrina Nykänen | Jouni Taskinen | Mahsa Hajisafarali | Anna Kuparinen Department of Biological and Environmental Science, University of Jyväskylä, Jyväskylä, Finland Correspondence Sabrina Nykänen, Department of Biological and Environmental Science, University of Jyväskylä, P.O. box 35 (YA), FI-40014 Jyväskylä, Finland. Email: [email protected] Funding information European Research Council, Grant/Award Number: 770884; Finnish Concordia Fund; Finnish Foundation for Nature Conservation; Kolarctic Cross Border Collaboration Programme, Grant/Award Number: KO1017; EU LIFE Programme, LIFE Revives, Grant/Award Number: LIFE20/NAT/FI/ 000611; Ellen and Artturi Nyyssönen Foundation; OLVI Foundation; Biological and Environmental Science Doctoral School of University of Jyväskylä Abstract Key life-history data, such as growth and age, are necessary to effectively manage and conserve threatened freshwater mussel species. Traditionally growth and age studies require large yet destructive sample sizes covering all age classes. Such methods pose a risk to populations of conservation concern, and therefore, alternative methods that need only limited sample sizes are necessitated to prevent further threats to such populations. We applied retrospective shell growth at age reconstructions to 98 critically endangered freshwater pearl mussel (FPM) individuals from 34 populations across Finland and Sweden, enabling the use of extremely small sample sizes (n=1–6 per population). We compared the performance of six different growth models with the reconstructed size-at-age data across FPM juvenile (<20 years old) and adult life stages. The growth reconstruction model showed reasonable skill in reconstructing FPM growth patterns. The von Bertalanffy model showed to be a good general descriptor of growth for FPM, but it systematically underestimated the asymptotic size. The power law model was the most accurate in estimating juvenile growth (lowest deviances from the size-at-age data). FPM showed great variability in longevity (A max =54–254 years) and growth constant k(0.018–0.057 year 1 ). Our results show that reasonable estimates of growth can be attained even when sample sizes are extremely limited. The results can be further applied to gain knowledge on the population's age structure, size at maturation, and recovery potential. The methodology is applicable to other freshwater mussel species of conservation concern. KEYWORDS age determination, back calculation, Bivalvia, endangered species, growth models, growth reconstruction, Unionida 1|INTRODUCTION Freshwater mussels (Unionida) are a diverse and widespread group of organisms, which have numerous important functional roles in freshwater ecosystems (Atkinson & Vaughn, 2014; DuBose et al., 2019). As filter feeders that often dominate benthic biomass, freshwater mussels contribute particularly to water purification, benthic-pelagic coupling, bioturbation and nutrient cycling (Howard & Received: 1 March 2024 Revised: 2 May 2024 Accepted: 14 May 2024 DOI: 10.1002/aqc.4205 This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes. © 2024 The Author(s). Aquatic Conservation: Marine and Freshwater Ecosystems published by John Wiley & Sons Ltd. Aquatic Conserv: Mar Freshw Ecosyst. 2024;34:e4205. wileyonlinelibrary.com/journal/aqc 1of15 https://doi.org/10.1002/aqc.4205 Cuffey, 2006; Strayer, 2014; Vaughn, 2018; Vaughn & Hakenkamp, 2001). However, due to anthropogenic activities, endemic freshwater mussels are highly imperilled and declining at some of the highest known rates worldwide, which has made them the focus of extensive conservation efforts (Geist et al., 2023; Lopes-Lima et al., 2017; Régnier et al., 2009; Strayer et al., 2004). Freshwater mussel species differ from each other especially in allocation to growth, which is reflected in great variation in growth parameters and longevity (Haag & Rypel, 2011). Somatic growth plays a key role in population dynamics and conservation biology because it strongly influences other central life-history traits, such as age at maturity, fecundity, survival and longevity (Charnov, 1993; Haag, 2013; Roff, 1992; Stearns, 1992). Understanding the characteristic of somatic growth is therefore fundamental in development of effective population conservation and management strategies for exploited or imperilled organisms (e.g., Ricker, 1975). While phylogeny constrains to some extent the life-history traits in freshwater mussels, within species growth and longevity show considerable plasticity in response to local environmental conditions, such as temperature and hydrochemistry (Bauer, 1992; Haag & Rypel, 2011; Jokela & Mutikainen, 1995). Because of the substantial plasticity in growth and longevity, generalizations of existing growth and age data from other species or even populations of the same species should be avoided as they can lead to wrong conclusions on the dynamics of a population of interest and ineffective or even harmful management and conservation strategies. Several mathematical functions have previously been used for studying the age-dependent growth patterns in freshwater mussels, but von Bertalanffy's (1938) growth function remains the most applied in literature (e.g., Haag & Rypel, 2011). Despite its wide use in growth studies for organisms expressing indeterminate growth (i.e., continuously through life), the von Bertalanffy growth function does not always perform well with growth in the youngest age classes (Gamito, 1998; Hastie et al., 2000; Miguel et al., 2004). Thus, to characterize adequately the growth of a given species or individual, it may be necessary to compare the performance of alternative models to von Bertalanffy. Further, traditionally growth models require sizeat-age data collected from multiple individuals of different sizes and ages from the same population to cover the growth trajectory from juvenile to adult because the models are highly dependent on having observations for all the age classes (Haag, 2009; Kritzer et al., 2001; Pardo et al., 2013). In the case of imperilled species, collection of live samples for growth and age studies is however often restricted because of the risk of destructive sampling. Helama and Valovirta (2008) addressed this problem by creating a model with which is possible to reconstruct the growth trajectories of freshwater mussels with internal annual growth increments determined in age studies. The model helps limiting destructive sampling in endangered populations in several ways. Firstly, it allows growth history reconstruction without mussel juveniles, which may be limited in unviable freshwater mussel populations. Second, reconstructing shell growth for size-at-age estimates makes it possible to conduct growth analyses without significantly reducing the number of reproductive individuals in the population as it reduces the number of live samples needed for growth analysis. Third, the model can also be applied to museum shell collections and to already existing growth data obtained from literature, making it an ethical technique of studying growth in endangered freshwater mussel species. In addition, the Helama and Valovirta (2008) model makes possible to produce growth trajectories and parameters at individual level, thus study the variation in growth within the populations of interest. Among Unionida, freshwater pearl mussel (FPM) Margaritifera margaritifera (Linnaeus, 1758) is a slow-growing and extremely long-lived species (Dunca et al., 2011; Haag & Rypel, 2011) that inhabits oligotrophic streams and rivers in the Holarctic region (Geist, 2010; Young et al., 2001). FPM is assessed as critically endangered and under threat of worldwide extinction without prompt conservation actions (Cuttelod et al., 2011; Lopes-Lima et al., 2017). Although the threats to FPM are manyfold (e.g., habitat deterioration, decline of host fish populations), low recruitment due to high juvenile mortality is considered the biggest factor affecting the decline of FPM populations (Geist, 2010; Österling et al., 2008). Studies on the age-dependent growth patterns of FPM in different populations can provide substantial information for effective conservation of the species. For example, growth patterns can be used to transform existing size distributions, measured as part of viability studies, into more accurate population specific age distributions. From conservation perspective, it is important to know the age structure of populations of interest as it highly influences the population's growth and recovery potential: different age groups have different reproductive capabilities and rates of mortality (Charnov, 1993; Roff, 1992; Stearns, 1992). In addition, it is important to study the size of juveniles in different populations, as the proportion of juvenile mussels (less than 20 and 50 mm in length) is one of the main criteria in judging viability of FPM populations, but currently the criteria does not take into account population specific size differences at juvenile age (Oulasvirta et al., 2017). Given that FPM is critically endangered and that sample collection for age and growth studies is limited, it represents an ideal candidate for the appliance of the previously presented shell growth reconstruction model for conservation purposes. The present study not only contributes to the conservation of one highly vulnerable species but also, more generally, demonstrates research practices that avoid destructive sampling while still gathering substantial life-history trait information —methodology that can be readily applied to other freshwater mussel species of conservation concern. In the present study we determined ages of 108 individuals and reconstructed the growth patterns of 98 individuals from 34 FPM populations across Finland and Sweden by applying the previously validated model by Helama and Valovirta (2008). We then compared the performance of several growth models in both juvenile and adult life stage. Therefore, the main aims of this study were to (1) reconstruct the growth patterns of mussel individuals using age determination data, (2) estimate the growth parameters and longevity of target populations, (3) investigate the among and within population variation in these parameters, (4) compare the performance of different growth models, and (5) find a suitable model for estimating the size of juveniles. 2of15 NYKÄNEN ET AL. 10990755, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aqc.4205 by University Of Jyväskylä Library, Wiley Online Library on [16/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 2|MATERIALS AND METHODS 2.1 |Study area and sampling In total, 29 rivers in northern Finland and 5 rivers in northern Sweden with resident FPM populations were sampled in the summer between 2019 and 2021 (June–September, Figure S1) as part of the European Neighbourhood Instrument Cross-Border Cooperation (ENI CBC) Kolarctic project ‘SALMUS’. Because of the endangered status of FPM, the collection of living individuals was performed with special permissions granted by the regional Centres for Economic Development, Transport and the Environment of Kainuu (KAIELY/296/2019 and 357/2019), North Ostrobothnia (POPELY/ 1276/2019 and 1490/2019) and Lapland (LAPELY/1929/2019 and 2252/2019) in Finland; and by the County Administrative Board of Norrbotten (623-7408-2020) in Sweden. Only rivers with population TABLE 1 Observed maximum shell length (L max , mm), shell height (H max , mm) and age (A max , years) for 29 Finnish and five Swedish freshwater pearl mussel (Margaritifera margaritifera) populations. Country Basin Catchment River nL max ,mm H max ,mm A max , years Finland Barents Sea River Teno Lovttajohka 3 120 57 254 Barents Sea River Tulomajoki Hanhioja 3 94 44 58 Barents Sea River Tulomajoki Kivijoki 3 98 54 134 Barents Sea River Tulomajoki Kolmosjoki 3 122 58 98 Barents Sea River Tulomajoki Lutto 3 (+3) 135 65 215 Barents Sea River Tulomajoki Nohkimaoja 3 102 45 a 120 Barents Sea River Tulomajoki Suomujoki 3 127 48 a 134 Barents Sea River Tulomajoki Torkojoki 3 110 46 a 113 Barents Sea River Tulomajoki Urakkajärvenoja 2 (+2) 111 46 a 110 Barents Sea River Tulomajoki Vuoksioja 1 95 41 a 54 Baltic Sea River Kemijoki Ahvenoja 3 120 54 93 Baltic Sea River Kemijoki Haukijoki 3 107 50 76 Baltic Sea River Kemijoki Satsijoki 3 102 45 97 Baltic Sea River Kemijoki Saukko-oja 3 101 43 107 Baltic Sea River Kemijoki Siikajoki 3 115 52 77 White Sea River Koutajoki Juumajoki 3 120 55 127 White Sea River Koutajoki Merenoja 3 140 64 150 White Sea River Koutajoki Myllyoja 3 103 48 100 White Sea River Koutajoki Porontimajoki 3 92 44 81 White Sea River Koutajoki Salmipuro 3 116 53 76 White Sea River Kem (Viena) Juomajoki 3 117 58 100 Baltic Sea River Iijoki Haukioja 3 110 50 80 Baltic Sea River Iijoki Livojoki 2 102 52 110 Baltic Sea River Iijoki Lohijoki 3 138 65 104 Baltic Sea River Iijoki Nuottipuro 3 104 48 100 Baltic Sea River Oulujoki Humalajoki 3 108 55 93 Baltic Sea River Oulujoki Mutajoki 3 83 41 88 Baltic Sea River Oulujoki Nuottijoki 3 (+5) 113 56 109 Baltic Sea River Oulujoki Varisjoki 3 117 52 125 Sweden Baltic Sea River Lule Souksaurebäcken 3 97 46 75 Baltic Sea River Lule Varjekbäcken 3 117 53 94 Baltic Sea River Pite Bölsmanån 3 131 59 92 Baltic Sea River Pite Ljusträskbäcken 3 115 52 61 Baltic Sea River Pite Tvättstugubäcken 3 91 42 72 Note: Number of age-determined living individuals is indicated by nand number of empty shells collected is in brackets. Total number of individuals retained for age determined was 108, of which in total 98 individuals were ≥50 years old and used in the shell reconstruction model. a Measured in laboratory from the shell cross-section produced for age determination. NYKÄNEN ET AL.3of15 10990755, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aqc.4205 by University Of Jyväskylä Library, Wiley Online Library on [16/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License size ≥1000 individuals were included in the present study. Within each river, in total 30 mussel individuals were sampled randomly from a representative location by snorkelling. All the mussels were measured with a vernier calliper for length and height. From the 30 individuals, three largest were retained (for exceptions, see Table 1) for age determination, making a total of 108 individuals. The rest of the random sample was returned alive to their collection sites. The retained individuals were immediately stored in ice and transported alive to the laboratory at the Department of Biological and Environmental Science, University of Jyväskylä, Finland. The mussel individuals were originally retained to determine their age from shells, but in addition, their tissues were individually stored accordingly for possible further analyses —such as DNA, stable isotopes, fatty acids, metals, parasites and morphology —the results of which are not presented here. 2.2 |Age determination and annual growth increments Age determination and measurement of the internal annual shell growth increments were performed in the Department of Palaeozoology, Swedish Museum of Natural History in Stockholm. Following the methods described by Dunca and Mutvei (2001), thin cross-sections were produced from one shell valve per mussel individual by cutting them perpendicularly to the annual growth rings (i.e., from the umbo to the marginal border). After being grinded and polished, the shell cross-sections were etched in Mutvei's solution for 30 min at 40C to improve the visibility of the winter lines and the precision of age estimations (Schöne et al., 2005). The shell crosssections were photographed with a reflective light microscope equipped with Carl Zeiss AxioCam camera. Using photographic enlargements, the internal annual growth increments were counted from the ventral margin to the beginning of the eroded part of the shell and measured to the nearest 1 μm as the minimum vertical distance between two winter lines in the prismatic (i.e., outer) shell layer, close to the nacreous layer's boundary line. The age of the eroded part in each shell was estimated with growth curves representing high, normal and low shell growth (Dunca et al., 2011). The age of the mussel was estimated by summing the number of the counted growth increments and age of the corroded part of the shell. For all populations, maximum observed age (A max , years), length (L max , mm) and height (H max , mm) were obtained. 2.3 |Reconstructing shell growth for height-at-age estimates We used a model developed by Helama and Valovirta (2008)to reconstruct the shell height growth history of each mussel individual throughout their life span. The model uses the observed annual shell growth increments measured from the shell cross-section to translate the convexly proceeding shell height growth (according to valve shape) into height-at-age estimates along the commissural plane — the direction from which the shell height is typically measured with venier callipers (see Figure 2in Helama & Valovirta, 2008). In Helama and Valovirta's (2008) model, the shell height (H t ) as a function of mussel age (t) can be reconstructed with the following equation: Ht¼Hcor þX t¼Amax t¼Acor þ1 htEtð1Þ where H t is the shell height (mm) at time t(age in years), H cor is the height (mm) of the corroded shell portion along the commissural plane, A max is the maximum observed age (i.e., age at time of death), A cor is the age of the corroded area (i.e., the number of missing increments), h t is the age-dependent adjustment factor for convexly occurring increment growth (fig. 5b in Helama & Valovirta, 2008), and E t is the external shell increment (mm, perpendicular to winter lines at the surface of the shell) at age t.H cor was measured in the lab with a microscope from the cross-sections. E t was estimated with the following equation: Et¼1 it Itð2Þ where i t describes the relationship between the internal and external shell increments as a function of age (fig. 4 in Helama & Valovirta, 2008), and I t is the internal shell increment (perpendicular to winter lines at the shell cross-section). As recommended by Helama and Valovirta (2008), only individuals of 50 years or older were used in the height-at-age reconstructions to avoid spurious fits, making a total of 98 mussel specimens in the subsequent growth model fitting process. The model-based terminal heights of the mussels were evaluated against the measured shell heights as model verification. 2.4 |Growth models Five nonlinear growth models, covering both juvenile and adult life stages, were fitted to the reconstructed shell height-at-age data to each individual mussel separately. The first was the von Bertalanffy (1938) growth function, which has primarily been applied in FPM and other bivalve age and growth studies (Haag & Rypel, 2011; Hastie et al., 2000; Miguel et al., 2004). The following 3-parameter von Bertalanffy growth function equation was used: Ht¼Hinf 1ektt0 ðÞ  ð3Þ where H t is the shell height (mm) at time t(age in years), H inf is the asymptotic height (mm), kis a growth constant that describes the rate at which H inf is reached (year 1 ) and t 0 is the theoretical age at which the height of the organism is zero (von Bertalanffy, 1938; Ricker, 1975). 4of15 NYKÄNEN ET AL. 10990755, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aqc.4205 by University Of Jyväskylä Library, Wiley Online Library on [16/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License The second was the Lester biphasic growth model, which is considered to perform well for species that mature relatively late and have long reproductive life spans (Lester et al., 2004). The Lester biphasic growth model consists of two separate equations for prematurity (all surplus energy invested to somatic growth) and postmaturity (surplus energy invested also to reproduction) periods (Lester et al., 2004). The growth in the pre-maturity period (t< age at maturity [T]) was modelled with a linear model fitted to immature size-at-age data: Ht¼htþcð4Þ where his juvenile growth rate (mm year 1 ) and cis the intercept of the linear fit to immature growth (mm). Twas set to 20 years as it is the expected age at maturity for FPM at our sampling latitudes (Arvidsson et al., 2012; Ziuganov et al., 1994). The x-intercept of the linear model is the hypothetical age at which mussel height is zero, denoted as t H=0 . We used h,t H=0 and Tin modelling the growth in the post-maturity period (t≥T) with mature size-at-age data and the following von Bertalanffy growth function, derived from Lester et al. (2004, eqs. 3.2, 3.3 and 3.4): Ht¼h=ek1  1ekTþln 1ek1 ðÞ TtH¼0 ðÞ ðÞ =k ðÞ t ðÞðÞ  ð5Þ where only the parameter kis unknown. The Lester biphasic growth model allows the amount of energy invested in reproduction to be estimated with the following expression from Lester et al. (2004, eq. 3.3): g¼3ek1  ð6Þ where g is the investment in reproduction (gonad weight/somatic weight). When gis solved, H inf and t 0 can be calculated with the equations from Lester et al. (2004, eqs. 3.2 and 3.4). In rare occasions, the pre-maturation growth could not be observed, and therefore, the Lester biphasic growth model could not be fitted. The third and fourth models were two sigmoidal growth (S-shaped) models. Both the Gompertz and logistic growth model have three parameters, but the former one is not symmetrical around the inflection point. We used self-starting model functions, which do not require initial values for the model parameter estimates. The Gompertz model was parameterized as follows: Ht¼Hinf ebct ðÞ ð7Þ where bis the displacement of the curve on the x-axis and cis the growth rate. The generalized logistic function was parameterized as follows: Ht¼Hinf =1þetmidtðÞ=scal  ð8Þ where t mid is the age when mussels have reached half of the asymptotic height (i.e., the inflection point), and scal is a scaling parameter for the x-axis with which the growth rate can be calculated as 1/scal. The fifth model was a power law model where the annual increase in shell height (δH t =H t+1 H t ) is equal to the mussel age raised to a power: δHt¼atbð9Þ where ais the absolute rate of growth and bis the rate at which the asymptotic size is approached. Under logarithmic transformation, the power law relationship become linearized: log10 δHt ðÞ¼log10 aðÞþblog10 tðÞ ð10Þ In order to estimate aand bin Equation (9), we fitted a log–log linear regression model to the data according to the Equation (10), where bis equivalent to the slope of the linear regression and ais the inverse-log transform of the y-intercept estimate. After solving aand b, the mussel height at certain age can be estimated with the integral of δH t with the following equation: Ht¼Hstart þð tn tstart atb:dt ð11Þ where H start is the height (mm) at the first observed annual increment, t start is the age (years) of the first observed increment, and t n is any given age (years). H inf was estimated by setting t n in Equation (11)to infinity (∞). In addition to the five models used to cover both juvenile and adult life stages, we additionally parameterized a sixth model for juvenile life stage only to test if the individual mussels followed exponential growth in the pre-maturity phase (up to the age of 19 years). We used an exponential growth model of the following form: Ht¼abtð12Þ where ais the initial model value H t when tis zero and bis the growth rate. Exponential growth model can be linearized using the natural logarithm of the response variable: ln Ht ðÞ¼ln aðÞþtln bðÞ ð13Þ We estimated aand bin Equation (12) with a log-level linear regression model according to the Equation (13), where aand bare the inverse-natural log transform of the y-intercept estimate and the slope of the linear regression model, respectively. After solving aand b, the size at given age can be estimated with the Equation (12). We did not have height observations between ages 0 to 7 years due to shell erosion. Therefore, the height of the FPM individuals at age 1 (H t=1 ) was set to 0.343 mm (n=57), which is an estimate of average shell height for FPM bred in captivity at the end of their NYKÄNEN ET AL.5of15 10990755, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aqc.4205 by University Of Jyväskylä Library, Wiley Online Library on [16/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License parasitic life stage (approximately 1 year old) (unpublished data). This approach facilitated the direction of the growth curves to biologically reasonable estimates. The additional data point were used in all models, except the power law model, because the latter require consecutive height observations. The parameters of all the models were estimated separately for each mussel individual. All the nonlinear models were fitted by nonlinear least-squares regression. The goodness of model fits to the height-at-age data were evaluated by visual inspection of the fitted lines and by comparing model regression coefficients (R 2 ), sums of squared residuals (RSS) and residual standard errors (RSE). Although the models have different model parameters, we could compare their associated estimates of asymptotic sizes. The models were compared with each other also with the deviances of the model height estimates from the observed shell heights, which were calculated subtracting the observed shell height at age tfrom the fitted shell height at age t. Statistical modelling was done using R (v. 4.2.1, R Core Team, 2022), the stats, the FSA (Ogle et al., 2022) and the minpack.lm packages (Elzhov et al., 2023). 3|RESULTS 3.1 |Maximum shell size and age The observed A max ,L max and H max values for each population are given in Table 1. The maximum life span showed remarkable variation across the populations, as the A max of the populations ranged from only 54 years (River Vuoksioja) to 254 years (River Lovttajohka). Variation can be seen also in the maximum shell sizes. The L max and H max ranged from 83 mm (River Mutajoki) to 140 mm (River Merenoja), and 41 mm (River Mutajoki) to 65 mm (River Lohijoki and River Lutto), respectively. Size, age and H cor of all mussel individuals can be found in the open data publishing platform Dryad (see Data availability statement). 3.2 |Reconstructed shell heights The measured shell heights indicated that mussels of similar age had different sizes, showing that the growth rate of FPM is highly variable across the study area (Figure 1a). The shell height growth reconstruction model showed reasonable skill in reconstructing the measured heights of the mussels. After accounting for outliers (Figure S2), linear regression showed low but statistically significant and positive relationship between the measured and reconstructed shell heights at the time of death (Figure 1b, adj. R 2 =0.401, Fstatistic =63.91, df =93, p< 0.001). The resulting growth trajectories from reconstructed shell heights of individual mussels revealed remarkable variation in size-at-age among the studied populations (Figure 2). For example, at 20 years of age (corresponding maturation; e.g., Ziuganov et al., 1994) the smallest and largest reconstructed heights were in average (± standard deviation) 14.2 mm (±1.0) in River Kivijoki and 39.9 mm (±4.0) in River Ljusträskbäcken, respectively (Figure 2). Variation in growth patterns can be seen also within populations, although is not as prominent as the among-river variation and our sample size per population was relatively small (n=1–6; Table 1and Figure 2). Visually inspecting, the shapes of the growth trajectories appear to be influenced by the age at time of death so that slower growth rates are associated with higher age at death (Figure S3). In addition, there is indication that only few individuals had reached the phase in their life span where the growth rate began to plateau, meaning that the growth was still significantly incomplete (Figure 2). 3.3 |Growth models' parameters and performance The growth models were applied to the individual shell growth history reconstructions presented in Figure 2. The von Bertalanffy growth model and logistic growth model converged for all the individuals. However, the Gompertz, Lester biphasic and power law growth models failed to converge 25, 16 and 3 times out of 98 individuals, respectively. There was some within-river variation in the performance of the growth models, but overall, the von Bertalanffy growth model was in average the model with the lowest RSS (113.1) and RSE (1.1), and highest R 2 (98.5%) (Table 2, Figure S4). The von Bertalanffy growth model outperformed the other models for 58.2%– 70.4% of the mussel individuals. For comparison, Gompertz and Lester biphasic growth models' relative performance was 21.4%– 25.5% and 4.1%–20.4%, respectively. The relative performance of the logistic and power law models was 0%. In addition, the power law model showed in average remarkably high RSS and RSE and low R 2 . This was likely caused by the fact that the power law model had one height observation less (i.e., H t=1 =0.343 mm) than the other models for each individual during the curve fitting process, but the goodnessof-fit estimators were calculated on the same number of residuals so that all estimators would be comparable. The visual inspection of the fitted growth curves showed that the von Bertalanffy growth model is generally a good growth model for FPM, but it also revealed that the model tends to underestimate the H inf (asymptotic height) of the mussel individuals (Figures S5 and S6). Nevertheless, Gompertz and logistic growth models systematically estimate even smaller H inf than the von Bertalanffy model. The number of cases where Gompertz growth model was the one fitting the best to a given individual increased slightly towards the southernmost catchments and decreased by age category (Figure S4). A closer look to the Gompertz model fits however reveals that, even if based on the RSS and RSE and R 2 values it outperformed the other models for some mussel individuals, in most of these cases the fitted Gompertz growth curves do not visually seem to fit the reconstructed height-at-age data better than the von Bertalanffy (Figure S5). The average of the von Bertalanffy growth model parameter estimates are listed in the Table 3. Average growth constant kestimates for the studied rivers ranged from 0.018 year 1 (River Hanhioja) to 0.057 year 1 (River Mutajoki and River Porontimajoki). 6of15 NYKÄNEN ET AL. 10990755, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aqc.4205 by University Of Jyväskylä Library, Wiley Online Library on [16/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Average estimates of H inf and t 0 ranged from 37.5 mm (River Saukkooja) to 79.5 mm (River Lohijoki) and from 6.5 years (River Lovttajohka) to 3.9 years (River Humalajoki), respectively. Data summarizing the average Gompertz, logistic, Lester biphasic, and power law growth model parameters can be found in the open data publishing platform Dryad (see Data availability statement). 3.4 |Deviance of the predicted sizes from the reconstructed sizes The deviances of the modelled shell heights from the reconstructed shell heights are illustrated in Figure 3a for the adult phase and in Figure 3b for the juvenile phase (see also Figure S7). 0 20 40 60 0 100 200 Age (years) Measured shell height ( mm ) 30 40 50 60 70 80 40 50 60 Measured shell height (mm) Reconstructed shell height ( mm ) (a) (b) FIGURE 1 (a) Measured shell heights plotted against the age of the freshwater pearl mussel (Margaritifera margaritifera) individuals (108 individuals from 34 populations). (b) Linear regression between the reconstructed shell heights and measured shell heights at the time of death (in total 98 individuals of age ≥50 years). Adj. R 2 =0.401, Fstatistic =63.91, df =93, RSE =6.46, p< 0.001. Black dots represent the data points, empty dots the outliers. Grey line is the regression line, and grey area is the 95% confidence region. Red line represents a line with slope of 1. NYKÄNEN ET AL.7of15 10990755, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aqc.4205 by University Of Jyväskylä Library, Wiley Online Library on [16/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Overall, most of the deviance in the height in the adult phase falls between 2.5 and 2.5 mm (Figure 3a). From all the models, the von Bertalanffy growth model has lowest deviance from the 0-line (i.e., no difference between the modelled and reconstructed height) and the model's distributions are narrower compared with the other growth models distributions. However, from 180 years age onwards all the models appear to underestimate the reconstructed heights, except for the power law model from approximately 200 years onwards. Figure 3b shows that in general the fitted juvenile (<20 years old) heights deviate more from the reconstructed heights than the fitted adult heights. However, size-and-age data for ages from 2 to 7 years were not obtained because of shell corrosion, and thus these ages' reconstructed heights could not be compared with the modelled heights. The age specific differences in the predicted and reconstructed heights show that the power law model can fit juvenile heights with the lowest deviance from the reconstructed heights when compared with the other growth models (Figure 3b). However, FIGURE 2 Reconstructed shell height-at-age of the freshwater pearl mussel (Margaritifera margaritifera) individuals by population (in total 98 individuals of age ≥50 years from 34 populations). Grey lines represent the individuals reconstructed growth history. Red lines represent the outlier individuals (see Figures 1b and S2). Dots indicate the measured heights of the mussels at the time of collection/death. 8of15 NYKÄNEN ET AL. 10990755, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aqc.4205 by University Of Jyväskylä Library, Wiley Online Library on [16/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Strayer, D.L., Downing, J.A., Haag, W.R., King, T.L., Layzer, J.B., Newton, T.J. et al. (2004). Changing perspectives on pearly mussels, North America's most imperiled animals. Bioscience, 54(5), 429–439. https:// doi.org/10.1641/0006-3568(2004)054[0429:CPOPMN]2.0.CO;2 Vaughn, C.C. (2018). Ecosystem services provided by freshwater mussels. Hydrobiologia, 810(1), 15–27. https://doi.org/10.1007/s10750-0173139-x Vaughn, C.C. & Hakenkamp, C.C. (2001). The functional role of burrowing bivalves in freshwater ecosystems. Freshwater Biology, 46(11), 1431– 1446. https://doi.org/10.1046/j.1365-2427.2001.00771.x Young, M.R., Cosgrove, P.J. & Hastie, L.C. (2001). The extent of, and causes for, the decline of a highly threatened naiad: Margaritifera margaritifera. In: Bauer, G. & Wächtler, K. (Eds.) Ecology and evolution of the freshwater mussels Unionoida. Berlin, Heidelberg: Springer, pp. 337–357. Ziuganov, V., Zotin, A., Nezlin, L. & Tretiakov, V. (1994). The freshwater pearl mussels and their relationships with salmonid fish. Moscow: VNIRO Publishing House. Ziuganov, V., Miguel, E. S., Neves, R. J., Longa, A., Fernández, C., Amaro, R., Beletsky, V., Popkovitch, E., Kaliuzhin, S., & Johnson, T. (2000). Life span variation of the freshwater pearl shell: A model species for testing longevity mechanisms in animals. AMBIO: A Journal of the Human Environment, 29(2), 102–105. https://doi.org/10.1579/00447447-29.2.102 SUPPORTING INFORMATION Additional supporting information can be found online in the Supporting Information section at the end of this article. How to cite this article: Nykänen, S., Taskinen, J., Hajisafarali, M. & Kuparinen, A. (2024). Growth and longevity of the endangered freshwater pearl mussel (Margaritifera margaritifera): Implications for conservation and management. Aquatic Conservation: Marine and Freshwater Ecosystems, 34(6), e4205. https://doi.org/10.1002/aqc.4205 NYKÄNEN ET AL.15 of 15 10990755, 2024, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/aqc.4205 by University Of Jyväskylä Library, Wiley Online Library on [16/06/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License