Winter is not coming : evaluating impacts of changing winter conditions on coregonine reproductive phenology

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 4.0 https://creativecommons.org/licenses/by/4.0/ Winter is not coming : evaluating impacts of changing winter conditions on coregonine reproductive phenology © EDP Sciences, 2024 Published version Stewart, Taylor R.; Karjalainen, Juha; Zucchetta, Matteo; Goulon, Chloé; Anneville, Orlane; Vinson, Mark R.; Wanzenböck, Josef; Stockwell, Jason D. Stewart, T. R., Karjalainen, J., Zucchetta, M., Goulon, C., Anneville, O., Vinson, M. R., Wanzenböck, J., & Stockwell, J. D. (2024). Winter is not coming : evaluating impacts of changing winter conditions on coregonine reproductive phenology. International Journal of Limnology, 60, Article 17. https://doi.org/10.1051/limn/2024014 2024 RESEARCH ARTICLE Winter is not coming: evaluating impacts of changing winter conditions on coregonine reproductive phenology Taylor R. Stewart 1,2,* , Juha Karjalainen 3 , Matteo Zucchetta 4 , Chloé Goulon 5 , Orlane Anneville 5 , Mark R. Vinson 6 , Josef Wanzenböck 7 and Jason D. Stockwell 2 1 Department of Biology, University of Vermont, Burlington, Vermont, USA 2 Rubenstein Ecosystem Science Laboratory, University of Vermont, Burlington, Vermont, USA 3 Department of Biological and Environmental Science, University of Jyväskylä, Jyväskylä, Finland 4 Institute of Polar Sciences, National Research Council, Venice, Italy 5 French National Research Institute for Agriculture, Food and the Environment (INRAE), University Savoie Mont Blanc, CARRTEL, Thonon-Les-Bains, France 6 United States Geological Survey, Great Lakes Science Center, Lake Superior Biological Station, Ashland, Wisconsin, USA 7 University of Innsbruck, Mondsee, Austria Received: 21 January 2024; Accepted: 5 July 2024 Abstract –Fishes in northern latitude lakes are at risk from climate-induced warming because the seasonality in water temperature is degrading, which can change ecosystem properties and the phenology of life-history events. Temperature-dependent embryo development models were developed for a group of cold, stenothermic fishes (Salmonidae Coregoninae) to assess the potential impacts of climate-induced changes in water temperature on cisco (Coregonus artedi) from two populations in Lake Superior (Apostle Islands [USA] and Thunder Bay [Canada]) and one in Lake Ontario (USA), vendace (C. albula) in Lake Southern Konnevesi (Finland), and European whitefish (C. lavaretus) in lakes Southern Konnevesi, Constance (Germany), Geneva (France), and Annecy (France). Water temperatures for each study group were simulated and changes in reproductive phenology across historic (1900–2006) and three future climatic-warming scenarios (2007–2099) were investigated. Models predicted that increases in water temperatures are likely to cause delayed spawning, shorter embryo incubation durations, and earlier larval hatching. Relative changes increased as warming scenarios increased in severity and were higher for littoral as compared to pelagic populations. Our simulations demonstrated that slower cooling in the autumn and (or) more rapid warming in spring can translate into substantial changes in the reproductive phenology of coregonines among our study groups. We expect that the changes in reproductive phenology predicted by our models, in the absence of thermal or behavioral adaptation, will have negative implications for population sustainability. Keywords: Coregonus / climate change / simulation modeling / reproductive phenology / water temperature 1 Introduction Warming lake temperatures worldwide (O’Reilly et al., 2015;Maberly et al., 2020;Woolway, 2023) are an imminent threat to lacustrine fish (Dahlke et al., 2020). Water temperature directly affects the limnological characteristics of lakes and is a master factor in regulating reproduction, development, physiology, and survival of lacustrine fishes (Gillooly et al., 2002;Brown et al., 2004;Cline et al., 2013; Little et al., 2020). Unlike riverine or marine fishes, lacustrine fish are limited in their ability to evade warming habitats. The vulnerability of fish populations varies among lakes depending on habitat connectivity, the magnitude of thermal regime change, and species and life-stage specific temperature tolerances (Dahlke et al., 2020). Holarctic fishes are predicted to be at risk because the strong seasonality in water temperature regimes in northern hemisphere lake ecosystems is degrading (Winslow et al., 2017;Sharma et al., 2019;Maberly et al., 2020;Woolway et al., 2021) and their life history is innately linked to seasonal climate patterns (Winder and Schindler, 2004). Disrupted seasonal patterns have resulted in positive feedback loops with warmer winter temperatures and (or) shorter periods of winter ice coverage, earlier and more rapid spring water warming, *Corresponding author: [email protected] Special issue - Biology and Management of Coregonid Fishes - 2023 Guest editors: Orlane Anneville, Chloé Goulon, Juha Karjalainen, Jean Guillard, Jared T. Myers and Jason Stockwell Int. J. Lim. 2024, 60, 17 ©EDP Sciences, 2024 https://doi.org/10.1051/limn/2024014 Available online at: www.limnology-journal.org Thisis an Open Access article distributed underthe terms ofthe Creative CommonsAttribution License (https://creativecommons.org/licenses/by/4.0), whichpermits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. warmer and prolonged summer growing seasons and stratification periods, and delayed autumn cooling (Sharma et al., 2019;Maberly et al., 2020;Woolway et al., 2021; Woolway, 2023). These changes in seasonal temperatures and lake mixing have modified lake ecosystem productivity (O'Reilly et al., 2015;Yankova et al., 2017), which can have large impacts on ecosystem structure and function (Carpenter et al., 2011;Bhagowati and Ahamad, 2019). Most observed freshwater fish responses to climateinduced changes in thermal conditions are shifts in withinlake habitat distributions (Tunney et al., 2014;Guzzo et al., 2017) and seasonal timing of life-history events (Parmesan, 2006;Farmer et al., 2015;Slesinger et al., 2021;Woods et al., 2021). Spawning adults and embryos are hypothesized to be the most temperature-sensitive life-stages in fishes based on the concept of limited cardiorespiratory capacity and thermal tolerance; spawning adults need suitable thermal habitat for offspring to develop and aerobic capacity improves with cardiorespiratory development (Dahlke et al., 2020). Warming waters will also shorten embryo incubation times resulting in earlier larval hatch dates (Reist et al., 2006;Karjalainen et al., 2015;Stewart et al., 2021a). Fish reproductive strategies have generally evolved to account for seasonal changes in resources and allow for the most energy-intensive period of the consumer's reproductive phenology to align with the peak availability of prey resources (Hjort, 1914;Cushing, 1990; Lowerre-Barbieri et al., 2011;Lyons et al., 2015). The spatiotemporal synchrony between larval fish and their prey is believed to be an important driver of interannual variation in fish year-class strength (Cushing, 1990;Nyberg et al., 2001; Straile et al., 2015). Freshwater coregonines, Salmonidae Coregoninae, are a group of cold, stenothermic fish species of high economic, ecological, and cultural importance widely distributed throughout the northern hemisphere (Bogue, 2001;Nyberg et al., 2001;Zeller et al., 2011;Lynch et al., 2016;Hodgson et al., 2020;Leppi et al., 2023). Over the past several decades, many coregonine populations have declined and are the focus of reintroduction, restoration, and conservation efforts (Zimmerman and Krueger, 2009;Anneville et al., 2015; Bronte et al., 2017). Declines in some coregonine populations throughout their range are related to low early-life survival (Nyberg et al., 2001;Parks and Rypel, 2018). Reasons for declining recruitment have been linked to invasive species (e.g., three-spined stickleback Gasterosteus aculeatus in Lake Constance) and changes in primary production (e.g., re-oligotrophication in Lake Geneva) at local scales (DeWeber et al., 2022;Bourinet et al., 2023). However, factors influencing declining recruitment remain unknown for many bodies of water, but climate-induced increases in water temperatures during critical early-life stages may be an important driver (Nyberg et al., 2001;Jeppesen et al., 2012; Karjalainen et al., 2015;Bourinet et al., 2023). Coregonines generally spawn near shore in late-autumn or winter (i.e., ca. October-January in the northern hemisphere); embryos incubate under ice or in ice-free water and hatch in late-winter or spring (Anneville et al., 2007;Karjalainen et al., 2015;Eshenroder et al., 2016). For autumn-spawning coregonines, the seasonal decrease in water temperature initiates spawning and winter water temperature is positively related to embryo development rate and negatively related to incubation duration (Colby and Brooke, 1973;Luczyński and Kirklewska, 1984;Gillet, 1989;Karjalainen et al., 2015; Eshenroder et al., 2016;Stewart et al., 2021a). Climate change projections suggest that warming lake temperatures in autumn and winter could alter spawning phenology, incubation time, and subsequent hatching time and survival of early-life stage coregonines (Nyberg et al., 2001;Karjalainen et al., 2015, 2016;Stewart et al., 2021b;Vinson et al., 2023). To further investigate how warming water temperatures may influence coregonine embryo development, we modeled the response of coregonine spawning phenology and development time to future climatic-warming scenarios for coregonine populations across North America and Europe. We hypothesized that delays in coregonine spawning, as a consequence of changes in autumn or winter cooling, result in altered embryo incubation durations (i.e., number of days between spawning and hatching) and hatching times. Our predictions were that delayed cooling in autumn or winter water temperature would cause delayed spawning times, with one of the following subsequent scenarios for hatching times: (1) shorter winter duration, higher winter water temperatures, decreased embryo incubation durations, and earlier hatching times; or (2) rapid winter cooling, normal winter water temperatures, typical incubation durations, and delayed hatching times. We expected coregonine populations that historically spawn at warmer temperatures (i.e., lower latitude populations) to have higher relative changes in reproductive phenology because of climate warming compared to populations that historically spawn at colder temperatures and are adapted to prolonged incubations (i.e., higher latitude populations). 2 Methods Temperature-dependent embryo development models were developed for eight coregonine study groups, newly fit models were validated with existing in situ observations, lake bottom thermal regimes were developed based on climatic-warming scenarios, and future simulations for study groups were run with validated models (Fig. 1). 2.1 Study locations and species Study groups included pelagic cisco (Coregonus artedi) from two populations in Lake Superior (Apostle Islands [United States of America (USA)] and Thunder Bay [Canada]) and one in Lake Ontario (USA), pelagic vendace (C. albula) in Lake Southern Konnevesi (Finland), and the littoral species within the European whitefish species complex (C. lavaretus) in lakes Southern Konnevesi, Geneva (France), and Annecy (France), and the pelagic species within the European whitefish species complex (C. lavaretus) in Lake Constance (Germany; Fig. 2). Study groups were selected based on the availability of water temperature, spawning, and developmental data. Lakes Superior and Ontario are large, deep lakes (>19,000 km 2 ; max depths >244 m) within the Laurentian Great Lakes system with a diverse range of trophic states, thermal conditions, and habitats. Lake Southern Konnevesi is an oligotrophic lake (120 km 2 ; max depth 56 m) in southern Finland with free-flowing inlet and outlet streams. Lakes Page 2 of 16 T.R. Stewart et al.: Int. J. Lim. 2024, 60, 17 Constance and Geneva are large, deep, peri-alpine lakes (>530 km 2 ; max depths >250 m) in central and western Europe. Lake Annecy is a small, peri-alpine lake (28 km 2 ; max depth 82 m) in western Europe. Lakes Constance, Geneva, and Annecy exhibit significant thermal stratification and have been shifting towards an oligotrophic state over the last 40 years (Bourinet et al., 2023). 2.2 Temperature-dependent embryo development model description Embryo development rates were expressed as the reciprocal number of days from fertilization to a given developmental stage (Colby and Brooke, 1973;Luczyński and Fig. 1. Workflow diagram describing the use of in situ spawning, hatching, and daily lakebed water temperature data in the model validation process. Fig. 2. Map showing the location of each study group (LSTB = Lake Superior Thunder Bay, LSAI = Lake Superior Apostle Islands; LOCB = Lake Ontario Chaumont Bay; LSK = Lake Southern Konnevesi; LC = Lake Constance; LG = Lake Geneva; LA = Lake Annecy) sampled in North America (left) and Europe (right). Page 3 of 16 T.R. Stewart et al.: Int. J. Lim. 2024, 60, 17 Kirklewska, 1984;Eckmann, 1987). The 50% hatching development stage was chosen to be comparable among study groups. The generalized equation relating rate of development to 50% hatching (DR 50 ) with respect to incubation temperature (x;°C) is DR50 ¼abxcx2 where a,b, and care polynomial coefficients and the semilog form is log10DR50 ¼log10aþxlog10bþx2log10c: DR 50 models for cisco from lakes Superior and Ontario and vendace from Lake Southern Konnevesi were fit with experimental incubation duration data on embryos incubated at constant water temperatures of ca. 2.0, 4.5, 7.0, and 9.0 °C (Stewart et al., 2021a;Fig. 3,Tab. 1). An additional model was fit for littoral spawning morphotypes of European whitefish from Lake Geneva using embryo incubation data. Winter temperatures at Lake Geneva in the littoral zone normally decline to only 5–7°C, so only two incubation temperatures were evaluated (contemporary at 7.0 °C and warmer winters at 9.0 °C). Thus, only a linear model was fit for European whitefish in Lake Geneva. The semilog form (i.e., log 10 DR 50 ) was fit to obtain model coefficients for each study group with experimental incubation data using the mean number of days from fertilization at each incubation temperature. Furthermore, published DR 50 models were used for the littoral and pelagic spawning species of European Whitefish (C. lavaretus macrophthalmus and C. lavaretus wartmanni, respectively) from Lake Constance, vendace from Lake Kosno (Poland), and cisco from Pickerel Lake (USA; Colby and Brooke, 1973; Luczyński and Kirklewska, 1984;Eckmann, 1987;Tab. 1). Hereafter, DR 50 models fit to each study group from experimental data (Stewart et al., 2021a, in review) are referred to as population-specific models while DR 50 models with published polynomial coefficients for species outside of our study groups (Colby and Brooke, 1973;Luczyński and Kirklewska, 1984;Eckmann, 1987) are species-specific models. To predict development time to 50% hatching, log 10 DR 50 was calculated for each daily mean temperature since spawning and the antilog10 taken to estimate the daily proportion of development. When the cumulative daily proportions of development equaled one, 50% hatching was assumed to have occurred. 2.3 Data sources Fishery-dependent spawning data were available for Lake Superior (Apostle Islands and Thunder Bay; Tab. A1). Commercial fishers from Lake Superior used gillnets to target spawning aggregations and recorded date, sex, maturity, and gonad condition from a subset of their catch (e.g.,first 10 fish; Yule et al., 2008). Fishery-independent spawning data collected during spawning surveys or broodstock collections were available for lakes Southern Konnevesi, Annecy, Geneva, and Ontario (Tab. A1). Spawning adults were collected using gillnets for lakes Southern Konnevesi, Geneva, and Annecy and trap nets for Lake Ontario at known spawning times and locations. Spawning adults collected were either assessed for gonad condition (i.e., green, ripe, spent) if actively spawning or were transported and held in hatchery ponds fed by lake water until spawning was observed. A daily distribution of spawning intensity was calculated, except for Lake Annecy where only peak spawning was reported. Daily number of ripe spawning individuals was reported for Lake Geneva, while daily gonad condition was reported for lakes Southern Konnevesi and Ontario. Repeat larval abundance data (i.e., ca. weekly) were gathered for each study group to estimate dates of hatching (Tab. A1). Larval collections were made by either a seine or ichthyoplankton net through ice, or along the surface or at a stratified depth when lakes were ice-free (Perrier et al., 2012; Karjalainen et al., 2019;Lucke et al., 2020;McKenna et al., 2020). All larvae collections began prior to the start of hatching or with low abundance so the date of first capture was assumed to be the start hatch date, except for the Lake Superior (Apostle Islands) study group. Larvae from Lake Superior (Apostle Islands) were sampled on fixed dates annually, presumably during or after peak hatching, and thus the hatch date of individual larvae were back-calculated from total lengths assuming an absolute growth rate of 0.18 mm day 1 (Oyadomari and Auer, 2007) and a length-at-hatch of 9.9 mm (Stewart et al., 2021a). Daily lakebed winter water temperatures were obtained from deployed temperature loggers or sondes (e.g., HOBO ® Water Temperature Pro v2) for lakes Superior (Apostle Islands), Ontario, Southern Konnevesi, Geneva, and Annecy on known spawning habitat or depth (Table A.1). Only a single winter (2017-18) of lakebed temperature data was available from Chaumont Bay, Lake Fig. 3. Predicted daily proportion of embryo development at water temperatures (°C) from validated development to 50% hatching (DR 50 )models for vendace (Coregonus albula) from Lake Southern Konnevesi, cisco (C. artedi) from lakes Superior and Ontario, and European whitefish (C. lavaretus) from Lake Geneva, and the pelagic species of European whitefish from Lake Constance. PopulationspecificDR 50 models were fit to data in Stewart et al., (2021a) for C. artedi from lakes Superior and Ontario and C. albula from Lake S. Konnevesi and Stewart et al., in review) for European whitefish from Lake Geneva. Species-specificDR 50 models were taken from Eckmann (1987) for C. lavaretus (pelagic) from Lake Constance. Page 4 of 16 T.R. Stewart et al.: Int. J. Lim. 2024, 60, 17 Ontario, and thus water temperatures from satellite surface water (https://podaac.ipl.nasa.gov/MEaSUREs-MUR) and U.S. Geological Survey Oswego River gauge (gauge number 04249000; U.S. Geological Survey, 2024) were compared to the observed lakebed winter water temperatures from Chaumont Bay in 2017-2018. The Oswego River gauge provided the closest representation of the lakebed thermal conditions of Chaumont Bay (mean daily difference = 0.77 °C) and was used as the in situ winter water temperatures for the Lake Ontario study group. Daily lakebed winter water temperatures for Lake Superior (Thunder Bay) were provided by the City of Thunder Bay Bare Point Water Treatment Plant, which measures water temperature at intake pipes submerged 733 m from shore at 13-m depth. 2.4 Model validation Population-specificDR 50 models fit from experimental incubation duration data (Stewart et al., in review) were validated using in situ daily lakebed winter water temperatures at typical spawning depths to compare model outputs (i.e., hatching dates) to the observed hatching data from eight coregonine study groups across North America and Europe. Experimental data were not available to fit population-specific models for lakes Annecy and Superior (Thunder Bay), thus the population-specific model from the closest-related study group and lake was used as the best-approximating model. Our models from Lake Geneva and the Apostle Islands in Lake Superior were therefore applied to Lake Annecy and Thunder Bay, respectively. Initiation and completion of contemporary spawning were based on observed water temperatures (Fig. 1). Coregonine spawning data to calculate the spawning temperatures were either (1) the daily proportion of ripe females leading to spawning or (2) the daily number of spawning individuals on spawning grounds. The onset of spawning was assumed to start on the date when either >10% of females captured daily were ripe or when the number of daily spawners was >10% of the total number of spawning individuals annually. The end of spawning was calculated as the nearest date to the start of spawning when either <10% of females were ripe or when the number of daily spawners was <10% of the total number of spawning individuals annually. If the end of spawning period could not be calculated (e.g., commercial roe fishery ceased once eggs were free flowing or only peak spawning was observed), the spawning period was assumed to end 9 days after the start of spawning (Straile et al., 2015). The water temperatures from the first and last date of the spawning period were taken from in situ daily lakebed winter water temperatures and averaged across all years available to define contemporary start and ending spawning water temperatures. If the years with spawning data did not overlap the years with water temperature data, the mean starting and ending days of year for the spawning period were calculated from all years of available spawning data, and the water temperatures at each mean starting and ending spawning days of year were averaged across all available water temperature years. For Lake Constance, where the dominant species of European whitefish spawns in pelagic water and embryos sink to a depth of 200– 250 m (Straile et al., 2007), spawning start dates were calculated using published equations from the date that the upper 10-m of the water column reached 10 °C isothermal conditions and spawning was assumed to end after 9 days (Straile et al., 2015). To validate the models for each study group, the mean spawning date was estimated from the calculated spawning start and end water temperatures for each year of available in situ water temperature data (Fig. 1). Mean observed hatching days of year were calculated as a weighted mean with daily larval abundance as the weight across all years available for each study group. The predicted development time to 50% hatching for each study group was calculated using daily water temperatures beginning at the mean spawning date for each population-specific model and, if applicable, the respective species-specific models from literature (Colby and Brooke, 1973;Luczyński and Kirklewska, 1984;Eckmann, 1987). The mean degree-days at hatching from all fitted models were Table 1. Development rate equations to 50% hatching DR 50 for cisco (Coregonus artedi from lakes Superior, Ontario, and Pickerel, vendace (C. albula) from lakes Kosno and Southern Konnevesi, and European whitefish (C. lavaretus) from lakes Constance and Geneva. PopulationspecificDR 50 models were fit to data for C. artedi from lakes Superior and Ontario, C. albula and C. lavaretus from Lake S. Konnevesi (Stewart et al., 2021a), and C. lavaretus from Lake Geneva (Stewart et al., in review). Species-specificDR 50 models were taken for C. artedi from Pickerel Lake (Colby and Brooke, 1973), C. albula from Lake Kosno (Luczyński and Kirklewska, 1984), and C. lavaretus (pelagic and littoral species) from Lake Constance (Eckmann, 1987). The coefficient of determination (R 2 ) was provided for models fit in this paper. All coefficients are in common logarithm (log10), and coefficients from Eckmann (1987) were transformed from natural logarithm. x is the incubation temperature in °C. –indicates no value was reported. Model group Species log 10 DR 50 =R 2 Lake Superior C. artedi 2.3289 + 0.0717x0.0001x 2 0.97 Lake Ontario C. artedi 2.3836 + 0.0643x+ 0.0008x 2 0.98 Pickerel Lake C. artedi 2.4088 + 0.0720x+ 0.0011x 2 – Lake Kosno C. albula 2.3035 + 0.0651x+ 0.0004x 2 – Lake S. Konnevesi C. albula 2.3664 + 0.0088x+ 0.0050x 2 0.94 Lake S. Konnevesi C. lavaretus 2.4183 + 0.0459x+ 0.0032x 2 0.96 Lake Constance C. lavaretus (pelagic) 2.2419 + 0.1104x0.0033x 2 – Lake Constance C. lavaretus (littoral) 2.3002 + 0.1104x0.0031x 2 – Lake Geneva C. lavaretus 2.1159 + 0.0528x0.86 Page 5 of 16 T.R. Stewart et al.: Int. J. Lim. 2024, 60, 17 estimated and compared to the mean observed degree-days at hatching for each study group. Degree-days (DD) were calculated using 0 °C as the reference temperature. A model was considered validated if the difference between the model degree-days at hatching and the mean observed degree-days at hatching was within ±50 DD. The closest validated model (i.e., either the population-specific model or species-specific model) to the mean observed DD at hatching was selected for each study group. 2.5 Climate scenarios Daily lake bottom water temperatures were simulated within the Inter-Sectoral Impact Model Intercomparison Project phase 2b (ISIMIP2b) local lake sector (Warszawski et al., 2014), using the SimStrat v2.1 lake model (Gaudard et al., 2019) for years 1900–2099 (Fig. A1). To drive the lake model, EWEMBI bias-corrected (Lange, 2019) climate model projections from ISIMIP2b were used, specifically projections from GFDL-ESM2M, HadGEM2-ES, IPSL-CM5A-LR, and MIROC5 for historic (1900–2006) and future periods (2007– 2099) under three representative concentration pathways (RCP): RCP 2.6, 6.0, and 8.5. These pathways included a range of potential future global radiative forcing with RCP 2.6 the lowest-emission scenario and RCP 8.5 the highestemission scenario (IPCC, 2014). To simulate how warming water temperatures would manifest in each of our study groups, lakes from ISIMIP were used as case studies. A total of 59 modeled case-study lakes from ISIMIP were categorized into climate zones following the definitions of Maberly et al. (2020) and maximum depth zones of 0–10, 10–25, 25–50, and 50+ m. Water temperatures from case-study lakes were averaged across all four climate model projections by each date, climate zone, and depth category to provide a simulated daily estimate of thermal spawning habitat for each climate zone and lake depth category combination. Each study group was assigned by climate zone and typical coregonine spawning depth, and the respective simulated mean daily water temperatures were used for each study group. Only 19 of the 59 available case-study lakes matched the climate zone and depth categories of our study groups and were used (Tab. A2). Lakes Geneva and Annecy are deep, peri-alpine lakes where coregonine spawning occurs in shallow waters (<5 m). The contemporary thermal spawning habitats for these two lakes were not well represented by the simulated bottom water temperature scenarios. Lakes Geneva and Annecy were available as modeled case-study lakes through ISIMIP and thus the lake-specific simulated daily water temperatures at the 5-m depth stratum were used, instead of the bottom water temperatures (mean maximum depth = 196 m), as the thermal scenarios for these lakes. 2.6 Model simulations For each simulated year in the model, spawning began on the first day that the 5-day running average of simulated water temperatures dropped below the population-specific start spawning temperature and ended on the first day that the 5-day running average of simulated water temperatures dropped below the population-specific end spawning temperature. Eggs were deposited throughout the defined spawning period according to the proportional rate of daily water temperature change, where larger daily decreases in temperature resulted in higher daily spawner abundances. We modeled 500 female spawners each year, with each female depositing 100 eggs to provide a relative index of cohort size. If the simulated water temperatures did not decrease sufficiently to end spawning, spawning was assumed to end after 20 days. A spawning period of this duration is reasonable for high-latitude coregonine populations (Karjalainen and Marjomäki, 2018). Daily cohorts of deposited eggs were run through the respective validated DR 50 model for each study group, and the hatch date was estimated. Table 2. Model validation results for population-specific development to 50% hatching models (DR 50 ) for cisco (Coregonus artedi) from lakes Superior (Apostle Island and Thunder Bay) and Ontario, vendace (C. albula) from Lake Southern Konnevesi, and European whitefish (C. lavaretus) from lakes Geneva and Annecy. Starting and ending spawning temperatures (°C) used as model parameters are listed and the model output (i.e., degree-days at hatching) from population-specific and species-specific models were compared to in situ hatching data for each study group. N indicates the number of available years with daily in situ water temperatures and across which spawning temperatures were averaged. Bolded degree-days at hatching indicates the selected DR 50 model for each study group, if one was validated. Population-specificDR 50 models were fit to data for C. artedi from lakes Superior (Apostle Islands) and Ontario, C. albula and C. lavaretus from Lake S. Konnevesi (Stewart et al., 2021a), and C. lavaretus from Lake Geneva (Stewart et al., in review). Species-specificDR 50 models were taken for C. artedi from Pickerel Lake (Colby and Brooke, 1973), C. albula from Lake Kosno (Luczyński and Kirklewska, 1984), and C. lavaretus (pelagic species) from Lake Constance (Eckmann, 1987). Mean spawning temperatures Mean degree-days at hatching Study group Species Start End Observed Pop. model Species model N Lake Superior (Apostle Islands) C. artedi 4.3 3.0 250.1 255.7 335.0 3 Lake Superior (Thunder Bay) C. artedi 4.9 4.4 452.3 337.9 405.3 4 Lake Ontario C. artedi 4.1 2.7 377.5 359.8 344.1 10 Lake S. Konnevesi C. albula 7.4 5.6 390.7 439.0 269.4 3 Lake S. Konnevesi C. lavaretus 6.0 4.7 292.5 395.7 200.6 3 Lake Geneva C. lavaretus 8.0 6.6 792.2 374.0 332.6 26 Lake Annecy C. lavaretus 7.9 6.4 342.5 370.3 332.3 16 Page 6 of 16 T.R. Stewart et al.: Int. J. Lim. 2024, 60, 17 To estimate the magnitude of change in reproductive phenology, incubation duration was calculated for each simulated embryo and mean spawning and hatching dates for each study group across historic (1900–2005) and future periods (2006–2099). Mean spawning date, incubation duration, and hatch date anomalies were calculated as deviations from the respective mean trait value during the historical period (1900–2005) for each study group, future year, and RCP scenario. Linear regressions were fit through simulated years to trait anomalies for each study group and RCP scenario. The linear models of the trait anomalies for each RCP scenario were compared using a two-way analysis of variance (R syntax: trait anomaly ~ year class + RCP scenario) within each study group, and a Tukey post-hoc test was conducted if a significant difference among slopes was found (a=0.05) using the emmeans package v.1.10.0 (Lenth, 2021). All simulations and analyses were performed in R version 4.2.3 (R Core Team, 2023). 3 Results Our population-specific models were fit, validated, and selected for cisco from lakes Superior (Apostle Islands) and Ontario, vendace from Lake Southern Konnevesi, and European whitefish from Lake Annecy (Tabs. 1 and 2). No model was validated for cisco from Lake Superior (Thunder Bay) and European whitefish from lakes Southern Konnevesi and Geneva because the difference between predicted and observed hatching was not within our confidence range (i.e., ±50 DD), and thus results are not reported for these study groups (Tab. 2). Study groups with validated DR 50 models displayed variable development rates but were similar within species across populations (Fig. 3). DR 50 models for European whitefish from Lake Constance had the fastest development rates while models for vendace from Lake Southern Konnevesi had the slowest development rates. 3.1 Spawning time Model simulations predicted that spawning will be delayed for all study groups except the pelagic species of European whitefish from Lake Constance as climate change scenarios increase water temperatures (Figs. 4 and 5). The RCP 8.5 scenario resulted in the greatest deviation from the mean historical spawning time in all study groups (mean increase = 0.28 days year 1 ). Spawning time of European whitefish from Lake Annecy had the largest response to increased temperature (0.61 days year 1 at RCP 8.5), and the RCP 8.5 scenario resulted in skipped spawning years and the complete absence of adequate Fig. 4. Simulation model anomalies (number of days) for spawning date, incubation length, and hatching date for cisco (Coregonus artedi) from lakes Superior and Ontario and vendace (C. albula) from Lake Southern Konnevesi. Anomalies were calculated for three representative concentration pathways (RCP) from 2007–2099 compared to the mean value from the historical period (1900-2006). The rate of simulated incubation temperature (°C) change per decade is indicated in the top row for each study group and matched to the respective RCP by color. Linear regression equations for each RCP scenario and the coefficient of determination (R 2 ) were provided. Page 7 of 16 T.R. Stewart et al.: Int. J. Lim. 2024, 60, 17 thermal habitat during the reproductive period required to initiate spawning by 2080 (Fig. 5). Cisco from lakes Superior and Ontario and vendace from Lake Southern Konnevesi had similar delayed spawning responses to increased temperatures within each RCP scenario (Fig. 4). Spawning time of European whitefish from Lake Constance responded similarly among RCP scenarios and had minimal change from historical spawning times (<0.07 days year 1 ;Fig. 5). All RCP anomaly slope pairwise comparisons were significantly different for spawning date within each study group (p<0.05), except for vendace from Lake Southern Konnevesi between RCP 2.6 and 6.0 (p = 0.060; Fig. 6). 3.2 Incubation length Embryo incubation durations (i.e., number of development days) were predicted to decrease because of increased water temperatures from climate change for all study groups examined (Figs. 4 and 5). The greatest deviation in incubation length from the mean historical incubation length was the RCP Fig. 5. Simulation model anomalies (number of days) for spawning date, incubation length, and hatching date for European whitefish (Coregonus lavaretus) from lakes Constance (pelagic species) and Annecy. Anomalies were calculated for three representative concentration pathways (RCP) from 2007-2099 compared to the mean value from the historical period (1900–2006). The rate of simulated incubation temperature ( °C) change per decade is indicated in the top row for each study group and matched to the respective RCP by color. Linear regression equations for each RCP scenario and the coefficient of determination (R 2 ) were provided. Page 8 of 16 T.R. Stewart et al.: Int. J. Lim. 2024, 60, 17 Leppi JC, Rinella DJ, Wipfli MS, Liljedahl AK, Seitz AC, Falke JA. 2023. Climate change risks to freshwater subsistence fisheries in Arctic Alaska: insights and uncertainty from broad whitefish Coregonus nasus. Fisheries 48: 295–306. Little AG, Loughland I, Seebacher F. 2020. What do warming waters mean for fish physiology and fisheries? J Fish Biol 97: 328–340. Lowerre-Barbieri SK, Ganias K, Saborido-Rey F, Murua H, Hunter JR. 2011. Reproductive timing in marine fishes: variability, temporal scales, and methods. Mar Coast Fish 3: 71–91. Lucke VS, Stewart TR, Vinson MR, Glase JD, Stockwell JD. 2020. Larval Coregonus spp. diets and zooplankton community patterns in the Apostle Islands, Lake Superior. J Gt Lakes Res 46: 1391–1401. Luczyński M, Kirklewska A. 1984. Dependence of Coregonus albula embryogenesis rate on the incubation temperature. Aquaculture 42: 43–55. Lynch AJ, Cooke SJ, Deines AM, Bower SD, Bunnell DB, Cowx IG, Nguyen VM, Nohner J, Phouthavong K, Riley B, Rogers MW, Taylor WW, et al., 2016. The social, economic, and environmental importance of inland fish and fisheries. Environ Rev 24: 115–121. Lyons J, Rypel AL, Rasmussen PW, Burzynski TE, Eggold BT, Myers JT, Paoli TJ, McIntyre PB. 2015. Trends in the reproductive phenology of two Great Lakes fishes. Trans Am Fish Soc 144: 1263–1274. Maberly SC, O'Donnell RA, Woolway RI, Cutler MEJ, Gong M, Jones ID, Merchant CJ, Miller CA, Politi E, Scott EM. 2020. Global lake thermal regions shift under climate change. Nat Commun 11: 1–9. Marjomäki TJ, Auvinen H, Helminen H, Huusko A, Sarvala J, Valkeajärvi P, Viljanen M, Karjalainen J. 2004. Spatial synchrony in the inter-annual population variation of vendace (Coregonus albula (L.)) in Finnish lakes. Ann Zool Fenn 41: 225–240. McKenna JE, Stott W, Chalupnicki M, Johnson JH. 2020. Spatial segregation of cisco (Coregonus artedi) and lake whitefish (C. clupeaformis) larvae in Chaumont Bay, Lake Ontario. J Gt Lakes Res 46: 1485–1490. McQuinn IH. 1997. Metapopulations and the Atlantic Herring. Rev Fish Biol Fish 7: 297–329. Miller TJ, Crowder LB, Rice JA, Marschall EA. 1988. Larval size and recruitment mechanisms in fishes: toward a conceptual framework. Can J Fish Aquat Sci 45: 1657–1670. Muir AM, Vecsei P, Pratt TC, Krueger CC, Power M, Reist JD. 2013. Ontogenetic shifts in morphology and resource use of cisco Coregonus artedi. J Fish Biol 82: 600–617. Myers JT, Yule DL, Jones ML, Quinlan HR, Berglund EK. 2014. Foraging and predation risk for larval cisco (Coregonus artedi) in Lake Superior: a modelling synthesis of empirical survey data. Ecol Model 294: 71–83. Nagahama Y, Yamashita M. 2008. Regulation of oocyte maturation in fish. Dev Growth Differ 50: S195–S219. Nyberg P, Bergstrand E, Degerman E, Enderlein O. 2001. Recruitment of pelagic fish in an unstable climate: studies in Sweden's four largest lakes. Ambio 30: 559–564. Ohlberger J, Mehner T, Staaks G, Hölker F. 2008. Is ecological segregation in a pair of sympatric coregonines supported by divergent feeding efficiencies? Can J Fish Aquat Sci 65: 2105– 2113. O'Reilly CM, Rowley RJ, Schneider P, Lenters JD, Mcintyre PB, Kraemer BM. 2015. Rapid and highly variable warming of lake surface waters around the globe. Geophys Res Lett 42: 1–9. Oyadomari JK, Auer NN. 2007. Influence of rearing temperature and feeding regime on Otolith increment deposition in larval Ciscoes. Trans Am Fish Soc 136: 766–777. Parks TP, Rypel AL. 2018. Predator-prey dynamics mediate longterm production trends of cisco (Coregonus artedi) in a northern Wisconsin lake. Can J Fish Aquat Sci 75: 1969–1976. Parmesan C. 2006. Ecological and evolutionary responses to recent climate change. Annu Rev Ecol Evol Syst 37: 637–669. Perrier C, Molinero JC, Gerdeaux D, Anneville O. 2012. Effects of temperature and food supply on the growth of whitefish Coregonus lavaretus larvae in an oligotrophic peri-alpine lake. J Fish Biol 81: 1501–1513. R Core Team, 2023. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing, https://www.r-project.org/. Reist JD, Wrona FJ, Prowse TD, Power M, Dempson JB, Beamish RJ, King JR, Carmichael TJ, Sawatzky CD. 2006. General effects of climate change on Arctic fishes and fish populations. AMBIO J Hum Environ 35: 370–380. Schulz M, Freyhof J, Saint-Laurent R, Østbye K, Mehner T, Bernatchez L. 2006. Evidence for independent origin of two spring-spawning ciscoes (Salmoniformes: Coregonidae) in Germany. J Fish Biol 68: 119–135. Sharma S, Blagrave K, Magnuson JJ, O'Reilly CM, Oliver S, Batt RD, Magee MR, Straile D, Weyhenmeyer GA, Winslow LA. 2019. Widespread loss of lake ice around the Northern Hemisphere in a warming world. Nat Clim Change 9: 227. Slesinger E, Jensen OP, Saba G. 2021. Spawning phenology of a rapidly shifting marine fish species throughout its range. ICES J Mar Sci 78: 1010–1022. Sommer U, Adrian R, De Senerpont Domis L, Elser JJ, Gaedke U, Ibelings B, Jeppesen E, Lürling M, Molinero JC, Mooij WM. 2012. Beyond the Plankton Ecology Group (PEG) model: mechanisms driving plankton succession. Annu RevEcol Evol Syst 43: 429–448. Stewart TR, Brun C, Goulon C, Baer J, Karjalainen J, Guillard J, Lasne E. in review. Response of whitefish embryos to thermal conditions diverges between perialpine populations. Int J Limnol. DOI: https://doi.org/10.1051/limn/2024017 Stewart TR, Mäkinen M, Goulon C, Guillard J, Marjomäki TJ, Lasne E, Karjalainen J, Stockwell JD. 2021a. Influence of warming temperatures on coregonine embryogenesis within and among species. Hydrobiologia 848: 4363–4385. Stewart TR, Vinson MR, Stockwell JD. 2021b. Shining a light on Laurentian Great Lakes cisco (Coregonus artedi): how ice coverage may impact embryonic development. J Gt Lakes Res 47: 1410–1418. Stockwell JD, Yule DL, Hrabik TR, Sierszen ME, Isaac EJ. 2014. Habitat coupling in a large lake system: delivery of an energy subsidy by an offshore planktivore to the nearshore zone of Lake Superior. Freshw Biol 59: 1197-1212. Straile D, Eckmann R, Jüngling T, Thomas G, Löffler H. 2007. Influence of climate variability on whitefish (Coregonus lavaretus) year-class strength in a deep, warm monomictic lake. Oecologia 151: 521–529. Straile D, Kerimoglu O, Peeters F. 2015. Trophic mismatch requires seasonal heterogeneity of warming. Ecology 96: 2794–2805. Tunney TD, McCann KS, Lester NP, Shuter BJ. 2014. Effects of differential habitat warming on complex communities. Proc Natl Acad Sci 111: 8077–8082. Page 15 of 16 T.R. Stewart et al.: Int. J. Lim. 2024, 60, 17 U.S. Geological Survey, 2024, USGS 04249000, Oswego River at Lock 7, Oswego, NY, in USGS water data for the Nation: U.S. Geological Survey National Water Information System database, accessed 14 September 2021 at https://doi.org/10.5066/F7P55KJN. [Site information directly accessible at https://waterdata.usgs.gov/ nwis/uv?site_no=04249000&legacy=1] Vinson MR, Herbert ME, Ackiss AS, Dobosenski JA, Evrard LM, Gorman OT, Lyons JF, Phillips SB, Yule DL. 2023. Lake Superior Kiyi reproductive biology. Trans Am Fish Soc 152: 75–93. Volta P, Jeppesen E, Sala P, Galafassi S, Foglini C, Puzzi C, Winfield IJ. 2018. Fish assemblages in deep Italian subalpine lakes: history and present status with an emphasis on non-native species. Hydrobiologia 824: 255–270. Warszawski L, Frieler K, Huber V, Piontek F, Serdeczny O, Schewe J. 2014. The inter-sectoral impact model intercomparison project (ISIMIP): project framework. Proc Natl Acad Sci 111:3228–3232. Winder M, Schindler DE. 2004. Climate change uncouples trophic interactions in an aquatic ecosystem. Ecology 85: 2100–2106. Winslow LA, Read JS, Hansen GJA, Rose KC, Robertson DM. 2017. Seasonality of change: summer warming rates do not fully represent effects of climate change on lake temperatures. Limnol Oceanogr 62: 2168–2178. Woods T, Kaz A, Giam X. 2021. Phenology in freshwaters: a review and recommendations for future research. Ecography 44: 1–14. Woolway RI. 2023. The pace of shifting seasons in lakes. Nat Commun 14: 2101. Woolway RI, Sharma S, Weyhenmeyer GA, Debolskiy A, Golub M, Mercado-Bettín D, Perroud M, Stepanenko V, Tan Z, Grant L. 2021. Phenological shifts in lake stratification under climate change. Nat Commun 12: 1–11. Yankova Y, Neuenschwander S, Köster O, Posch T. 2017. Abrupt stop of deep water turnover with lake warming: drastic consequences for algal primary producers. Sci Rep 7: 1–9. Yule DL, Stockwell JD, Black JA, Cullis KI, Cholwek GA, Myers JT. 2008. How systematic age underestimation can impede understanding of fish population dynamics: lessons learned from a lake superior Cisco Stock. Trans Am Fish Soc 137: 481–495. Zeller D, Booth S, Pakhomov E, Swartz W, Pauly D. 2011. Arctic fisheries catches in Russia, USA, and Canada: baselines for neglected ecosystems. Polar Biol 34: 955–973. Zimmerman MS, Krueger CC. 2009. An ecosystem perspective on reestablishing native deepwater fishes in the Laurentian Great Lakes. North Am J Fish Manag 29: 1352–1371. Cite this article as: Stewart TR, Karjalainen J, Zucchetta M, Goulon C, Anneville O, Vinson MR, Wanzenböck J, Stockwell JD. 2024. Winter is not coming: evaluating impacts of changing winter conditions on coregonine reproductive phenology. Int. J. Lim. 60: 17. Page 16 of 16 T.R. Stewart et al.: Int. J. Lim. 2024, 60, 17