Joanna Mills Flemming
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Joanna Mills Flemming has a total of 57 co-authors. Here are the top co-authors by number of shared publications:
- Daniel E. Ruzzante
Department of Biology Dalhousie University Halifax Nova Scotia Canada
4 shared publications Chris Field
Department of Mathematics & Statistics, Dalhousie University
3 shared publications- Natasha J. Klappstein
Department of Mathematics and Statistics Dalhousie University Halifax Nova Scotia Canada
3 shared publications - Cornelia Elizabeth Den Heyer
Fisheries and Oceans Canada, Bedford Institute of Oceanography, Dartmouth, Canada
2 shared publications - David M. Keith
Affiliation Fisheries and Oceans Canada, Bedford Institute of Oceanography, Dartmouth, Nova Scotia, Canada
2 shared publications - Ellen M. Weise
Department of Biology Dalhousie University Halifax Nova Scotia Canada
2 shared publications
Latest Publications
- Open Access
- 10 November 2026
- Published by Wiley in Molecular Ecology Resources
- Vol. 26 (8)
Non‐Linear Epigenetic Aging Clocks: A Case Study in Atlantic Halibut ( Hippoglossus hippoglossus )
Age information is fundamental in population biology. In fisheries management, robust and effective stock assessment models depend on accurate estimates of fecundity, survival rates, and other life‐history traits, many of which vary with age. Consequently, reliable age determination is essential for understanding population dynamics and supporting informed management and conservation decisions. Age in fish is generally estimated by counting otolith growth rings, a labor‐ and time‐intensive approach that requires lethal sampling and highly specialized, species‐specific expertise. To supplement current otolith‐based aging efforts, we developed a DNA methylation approach for aging Atlantic Halibut. We conducted whole‐genome methylation sequencing on 106 wild‐caught individuals with otolith‐derived age estimates, ranging from 4 to 29. CpG loci were evaluated as candidate age predictors using elastic net regression and a generalized additive model, with the latter allowing for the incorporation of non‐linear relationships between CpG methylation and age. Age‐associated loci identified by elastic net regression were primarily located in likely subtelomeric and intronic regions. Importantly, inclusion of older individuals in the data set led to changes in the selected CpG sites but not in their respective chromosomal regions, suggesting that it is the methylation patterns in the chromosomes, rather than specific CpG sites, that correlate with age. The sex‐separated generalized additive model (GAM) performed better than the elastic net regression as it exhibited the lowest per‐age median error (MAE < 1 year, or < 2% of the average Atlantic Halibut lifespan). Our framework for epigenetic clock calibration allows for robust, system‐specific clock calibration, which can be used for other non‐model organisms under exploitation and conservation concern.
- 10 November 2026
- Published by Wiley in Molecular Ecology Resources
- Vol. 26 (8)
Open AccessNon‐Linear Epigenetic Aging Clocks: A Case Study in Atlantic Halibut ( Hippoglossus hippoglossus )
Age information is fundamental in population biology. In fisheries management, robust and effective stock assessment models depend on accurate estimates of fecundity, survival rates, and other life‐history traits, many of which vary with age. Consequently, reliable age determination is essential for understanding population dynamics and supporting informed management and conservation decisions. Age in fish is generally estimated by counting otolith growth rings, a labor‐ and time‐intensive approach that requires lethal sampling and highly specialized, species‐specific expertise. To supplement current otolith‐based aging efforts, we developed a DNA methylation approach for aging Atlantic Halibut. We conducted whole‐genome methylation sequencing on 106 wild‐caught individuals with otolith‐derived age estimates, ranging from 4 to 29. CpG loci were evaluated as candidate age predictors using elastic net regression and a generalized additive model, with the latter allowing for the incorporation of non‐linear relationships between CpG methylation and age. Age‐associated loci identified by elastic net regression were primarily located in likely subtelomeric and intronic regions. Importantly, inclusion of older individuals in the data set led to changes in the selected CpG sites but not in their respective chromosomal regions, suggesting that it is the methylation patterns in the chromosomes, rather than specific CpG sites, that correlate with age. The sex‐separated generalized additive model (GAM) performed better than the elastic net regression as it exhibited the lowest per‐age median error (MAE < 1 year, or < 2% of the average Atlantic Halibut lifespan). Our framework for epigenetic clock calibration allows for robust, system‐specific clock calibration, which can be used for other non‐model organisms under exploitation and conservation concern.
- Open Access
- 1 August 2026
- Published by Acoustical Society of America (ASA) in JASA Express Letters
- Vol. 6 (8)
Estimating the minimum numbers of calling mysticetes using passive acoustic bearings
Monitoring the abundance and distribution of marine mammals is essential for evaluating human and environmental effects. Passive acoustic monitoring offers a non-visual alternative, but estimating animal numbers remains difficult. We introduce Spatial Counting of Animal Numbers (SCAN), a method that counts unique bearings of detected calls to estimate a conservative minimum number of animals present. SCAN is a complementary metric to traditional presence/absence results and calibrated density estimates. The method is demonstrated using sei whale calls recorded by a four-element array on an autonomous glider. SCAN increases the information on marine mammal populations derived from passive acoustic recordings.
- 1 August 2026
- Published by Acoustical Society of America (ASA) in JASA Express Letters
- Vol. 6 (8)
Open AccessEstimating the minimum numbers of calling mysticetes using passive acoustic bearings
Monitoring the abundance and distribution of marine mammals is essential for evaluating human and environmental effects. Passive acoustic monitoring offers a non-visual alternative, but estimating animal numbers remains difficult. We introduce Spatial Counting of Animal Numbers (SCAN), a method that counts unique bearings of detected calls to estimate a conservative minimum number of animals present. SCAN is a complementary metric to traditional presence/absence results and calibrated density estimates. The method is demonstrated using sei whale calls recorded by a four-element array on an autonomous glider. SCAN increases the information on marine mammal populations derived from passive acoustic recordings.
- Open Access
- 1 July 2026
- Published by Wiley in Ecology Letters
- Vol. 29 (7) , e70414
Three‐Dimensional Correlated Random Walks for Animal Movement and Habitat Selection
Animal movement and habitat selection underpin important ecological phenomena, from individual behaviour to population‐level distributions. Despite navigating three‐dimensional space, animals' movement is typically measured and analysed on a two‐dimensional plane, which limits our understanding of species that swim or fly. Therefore, we propose a step selection function (SSF) capable of quantifying animal movement and habitat selection in three dimensions. We formulate a very general family of three‐dimensional correlated random walks, aimed at capturing unique features of three‐dimensional data. Using Antarctic petrel data, we illustrate how these SSFs can be used to assess selection for vertically‐stratified habitat, account for barriers (e.g., the ground or ocean surface), and model attraction to any number of directional targets. Our modelling framework provides a solid foundation for three‐dimensional analyses, which will be crucial to answer ecological questions that would otherwise be ignored in two dimensions.
- 1 July 2026
- Published by Wiley in Ecology Letters
- Vol. 29 (7) , e70414
Open AccessThree‐Dimensional Correlated Random Walks for Animal Movement and Habitat Selection
Animal movement and habitat selection underpin important ecological phenomena, from individual behaviour to population‐level distributions. Despite navigating three‐dimensional space, animals' movement is typically measured and analysed on a two‐dimensional plane, which limits our understanding of species that swim or fly. Therefore, we propose a step selection function (SSF) capable of quantifying animal movement and habitat selection in three dimensions. We formulate a very general family of three‐dimensional correlated random walks, aimed at capturing unique features of three‐dimensional data. Using Antarctic petrel data, we illustrate how these SSFs can be used to assess selection for vertically‐stratified habitat, account for barriers (e.g., the ground or ocean surface), and model attraction to any number of directional targets. Our modelling framework provides a solid foundation for three‐dimensional analyses, which will be crucial to answer ecological questions that would otherwise be ignored in two dimensions.
- Open Access
- 1 January 2026
- Published by Canadian Science Publishing in Canadian Journal of Fisheries and Aquatic Sciences
- Vol. 83, 1-14
Incorporating image-based survey data within stock assessment models
Many new technologies are becoming available to assist with the assessment of fish stocks. These novel datasets have tremendous potential that has not yet been fully explored. One such technology is the drop camera, an underwater camera system lowered into the water column from a vessel to capture video or still images of the seafloor. We develop here straightforward methods for the inclusion of drop camera data within a spatio-temporal stock assessment model, and explore associated benefits via simulations. Results show that comparing drop-camera data and traditional dredge survey data can help estimate dredge survey catchability, while also identifying conditions under which each dataset outperforms the other. A case study on the sea scallop population on Georges Bank highlights this newfound ability to estimate catchability parameters, and demonstrates how to include both datasets within the same statistical framework. This research reveals that incorporating new types of data can substantially improve stock assessment models, but cannot fully replace traditional dredge surveys due to the inability of these new technologies to provide critical data such as weight measurements.
- 1 January 2026
- Published by Canadian Science Publishing in Canadian Journal of Fisheries and Aquatic Sciences
- Vol. 83, 1-14
Open AccessIncorporating image-based survey data within stock assessment models
Many new technologies are becoming available to assist with the assessment of fish stocks. These novel datasets have tremendous potential that has not yet been fully explored. One such technology is the drop camera, an underwater camera system lowered into the water column from a vessel to capture video or still images of the seafloor. We develop here straightforward methods for the inclusion of drop camera data within a spatio-temporal stock assessment model, and explore associated benefits via simulations. Results show that comparing drop-camera data and traditional dredge survey data can help estimate dredge survey catchability, while also identifying conditions under which each dataset outperforms the other. A case study on the sea scallop population on Georges Bank highlights this newfound ability to estimate catchability parameters, and demonstrates how to include both datasets within the same statistical framework. This research reveals that incorporating new types of data can substantially improve stock assessment models, but cannot fully replace traditional dredge surveys due to the inability of these new technologies to provide critical data such as weight measurements.
- Open Access
- 24 November 2024
- Published by Wiley in Molecular Ecology Resources
- Vol. 25 (3) , e14047
Temporal Variability in Effective Size (N̂e) Identifies Potential Sources of Discrepancies Between Mark Recapture and Close Kin Mark Recapture Estimates of Population Abundance
Although efforts to estimate effective population size, census size and their ratio in wild populations are expanding, few empirical studies investigate interannual changes in these parameters. Hence, we do not know how repeatable or representative many estimates may be. Answering this question requires studies of long‐term population dynamics. Here we took advantage of a rich dataset of seven brook trout (Salvelinus fontinalis) populations, 5 consecutive years and 5400 individuals genotyped at 33 microsatellites to examine variation in estimates of effective and census size and in their ratio. We first estimated the annual effective number of breeders (b) using individuals aged 1+. We then adjusted these estimates using two life history traits, to obtain and subsequently, following Waples et al. (2013). was estimated for the years 2014 to 2019. Census size was estimated by mark recapture using double‐pass electrofishing () (years 2014–2018) as well as by the Close Kin Mark Recapture approach () (years 2015–2017). Within populations, annual variation in (ratio of maximum to minimum ) ranged from 1.6‐fold to 58‐fold. Over all 7 populations, the median annual variation in was around 5‐fold. These results reflect important interannual changes in the variance in reproductive success and more generally in population dynamics. Within population varied between years by a (median) factor of 2.7 with a range from 2 to 4.3. Thus, estimated effective size varied nearly twice as much as did estimated census size. Our results therefore suggest that, at least in small populations like those examined in the present study, any single annual estimate of is unlikely to be representative of long‐term dynamics. At least 3–4 annual estimates may be required for an estimate of contemporary effective size to be truly representative. We then compared to . For five of the seven populations, the estimates of population abundance based on mark recapture () were indistinguishable from those based on close kin mark recapture (). The two populations with discordant and exhibited extremely low ratios and the largest annual variation in (58‐fold and 35.4‐fold respectively). These results suggest that sampling effort in these two streams may have been insufficient to properly capture the genetic diversity of the entire population and that individuals sampled were not representative of the population. Our study further validates CKMR as a method for estimating abundance in wild populations and it demonstrates how knowledge of the temporal variation in can be used to identify potential sources of discrepancies between and .
- 24 November 2024
- Published by Wiley in Molecular Ecology Resources
- Vol. 25 (3) , e14047
Open AccessTemporal Variability in Effective Size (N̂e) Identifies Potential Sources of Discrepancies Between Mark Recapture and Close Kin Mark Recapture Estimates of Population Abundance
Although efforts to estimate effective population size, census size and their ratio in wild populations are expanding, few empirical studies investigate interannual changes in these parameters. Hence, we do not know how repeatable or representative many estimates may be. Answering this question requires studies of long‐term population dynamics. Here we took advantage of a rich dataset of seven brook trout (Salvelinus fontinalis) populations, 5 consecutive years and 5400 individuals genotyped at 33 microsatellites to examine variation in estimates of effective and census size and in their ratio. We first estimated the annual effective number of breeders (b) using individuals aged 1+. We then adjusted these estimates using two life history traits, to obtain and subsequently, following Waples et al. (2013). was estimated for the years 2014 to 2019. Census size was estimated by mark recapture using double‐pass electrofishing () (years 2014–2018) as well as by the Close Kin Mark Recapture approach () (years 2015–2017). Within populations, annual variation in (ratio of maximum to minimum ) ranged from 1.6‐fold to 58‐fold. Over all 7 populations, the median annual variation in was around 5‐fold. These results reflect important interannual changes in the variance in reproductive success and more generally in population dynamics. Within population varied between years by a (median) factor of 2.7 with a range from 2 to 4.3. Thus, estimated effective size varied nearly twice as much as did estimated census size. Our results therefore suggest that, at least in small populations like those examined in the present study, any single annual estimate of is unlikely to be representative of long‐term dynamics. At least 3–4 annual estimates may be required for an estimate of contemporary effective size to be truly representative. We then compared to . For five of the seven populations, the estimates of population abundance based on mark recapture () were indistinguishable from those based on close kin mark recapture (). The two populations with discordant and exhibited extremely low ratios and the largest annual variation in (58‐fold and 35.4‐fold respectively). These results suggest that sampling effort in these two streams may have been insufficient to properly capture the genetic diversity of the entire population and that individuals sampled were not representative of the population. Our study further validates CKMR as a method for estimating abundance in wild populations and it demonstrates how knowledge of the temporal variation in can be used to identify potential sources of discrepancies between and .
- Open Access
- 18 September 2024
- Published by Wiley in Methods in Ecology and Evolution
- Vol. 15 (11) , 2059-2073
Estimating effective population size using close‐kin mark–recapture
Close‐kin mark–recapture (CKMR) is a method that allows estimating population census size, among other parameters, through the observed number of pairs that are close‐kin including parent–offspring pairs (POPs) and half‐sibling pairs (HSPs). CKMR models are capable of estimating abundance, fecundity and survival at age using POPs and HSPs from different cohorts. The link between effective population size (Ne) and the number of ‘within‐cohort’ sibling pairs has been noted before but how to actually achieve such an estimate with CKMR has not been previously demonstrated. We show it is possible to use the number of ‘within‐cohort’ sibling pairs along with POPs and ‘different‐cohort’ HSPs to estimate Ne. These can be combined with an estimate of annual variance in number of offspring to estimate the lifetime variance of total reproductive success which can be used to find Ne. We show that the variance in number of offspring produced by adults in a given year is related to the within‐cohort comparisons. Our approach is demonstrated on an individual‐based simulation where we show that the CKMR Ne estimate offers similar results to Ne estimated from methods like linkage disequilibrium. Our methods allow estimating Ne using CKMR while also estimating demographic parameters.
- 18 September 2024
- Published by Wiley in Methods in Ecology and Evolution
- Vol. 15 (11) , 2059-2073
Open AccessEstimating effective population size using close‐kin mark–recapture
Close‐kin mark–recapture (CKMR) is a method that allows estimating population census size, among other parameters, through the observed number of pairs that are close‐kin including parent–offspring pairs (POPs) and half‐sibling pairs (HSPs). CKMR models are capable of estimating abundance, fecundity and survival at age using POPs and HSPs from different cohorts. The link between effective population size (Ne) and the number of ‘within‐cohort’ sibling pairs has been noted before but how to actually achieve such an estimate with CKMR has not been previously demonstrated. We show it is possible to use the number of ‘within‐cohort’ sibling pairs along with POPs and ‘different‐cohort’ HSPs to estimate Ne. These can be combined with an estimate of annual variance in number of offspring to estimate the lifetime variance of total reproductive success which can be used to find Ne. We show that the variance in number of offspring produced by adults in a given year is related to the within‐cohort comparisons. Our approach is demonstrated on an individual‐based simulation where we show that the CKMR Ne estimate offers similar results to Ne estimated from methods like linkage disequilibrium. Our methods allow estimating Ne using CKMR while also estimating demographic parameters.
- Open Access
- 24 June 2024
- Published by Wiley in Methods in Ecology and Evolution
- Vol. 15 (8) , 1332-1346
Step selection functions with non‐linear and random effects
Step selection functions (SSFs) are used to jointly describe animal movement patterns and habitat preferences. Recent work has extended this framework to model inter‐individual differences, account for unexplained structure in animals' space use and capture temporally varying patterns of movement and habitat selection. In this paper, we formulate SSFs with penalised smooths (similar to generalised additive models) to unify new and existing extensions, and conveniently implement the models in the popular, open‐source mgcv R package. We explore non‐linear patterns of movement and habitat selection, and use the equivalence between penalised smoothing splines and random effects to implement individual‐level and spatial random effects. This framework can also be used to fit varying‐coefficient models to account for temporally or spatially heterogeneous patterns of selection (e.g. resulting from behavioural variation), or any other non‐linear interactions between drivers of the animal's movement decisions. We provide the necessary technical details to understand several key special cases of smooths and their implementation in mgcv, showcase the ecological relevance using two illustrative examples and provide R code to facilitate the adoption of these methods. This paper offers a broad overview of how smooth effects can be applied to increase the flexibility and biological realism of SSFs.
- 24 June 2024
- Published by Wiley in Methods in Ecology and Evolution
- Vol. 15 (8) , 1332-1346
Open AccessStep selection functions with non‐linear and random effects
Step selection functions (SSFs) are used to jointly describe animal movement patterns and habitat preferences. Recent work has extended this framework to model inter‐individual differences, account for unexplained structure in animals' space use and capture temporally varying patterns of movement and habitat selection. In this paper, we formulate SSFs with penalised smooths (similar to generalised additive models) to unify new and existing extensions, and conveniently implement the models in the popular, open‐source mgcv R package. We explore non‐linear patterns of movement and habitat selection, and use the equivalence between penalised smoothing splines and random effects to implement individual‐level and spatial random effects. This framework can also be used to fit varying‐coefficient models to account for temporally or spatially heterogeneous patterns of selection (e.g. resulting from behavioural variation), or any other non‐linear interactions between drivers of the animal's movement decisions. We provide the necessary technical details to understand several key special cases of smooths and their implementation in mgcv, showcase the ecological relevance using two illustrative examples and provide R code to facilitate the adoption of these methods. This paper offers a broad overview of how smooth effects can be applied to increase the flexibility and biological realism of SSFs.
- Open Access
Step selection analysis with non-linear and random effects in mgcv
Step selection analysis is used to jointly describe animal movement patterns and habitat preferences. Recent work has extended this framework to model inter-individual differences, account for unexplained structure in animals’ space use, and capture temporally-varying patterns of movement and habitat selection.In this paper, we formulate step selection functions with penalised smooths (similar to generalised additive models) to unify new and existing extensions, and conveniently implement the models in the popular, open-source mgcv R package. We explore non-linear patterns of movement and habitat selection, and use the equivalence between penalised smoothing splines and random effects to implement individual-level and spatial random effects. This framework can also be used to fit varying-coefficient models to account for temporally or spatially-heterogeneous patterns of selection (e.g., resulting from behavioural variation), or any other non-linear interactions between drivers of the animal’s movement decisions. We provide the necessary technical details to understand several key special cases of smooths and their implementation in mgcv , showcase the ecological relevance using two illustrative examples, and provide R code (available at https://github.com/NJKlappstein/smoothSSF ) to facilitate the adoption of these methods. This paper is a broad overview of how smooth effects can be applied to increase the flexibility and biological realism of step selection analysis.
Open AccessStep selection analysis with non-linear and random effects in mgcv
Step selection analysis is used to jointly describe animal movement patterns and habitat preferences. Recent work has extended this framework to model inter-individual differences, account for unexplained structure in animals’ space use, and capture temporally-varying patterns of movement and habitat selection.In this paper, we formulate step selection functions with penalised smooths (similar to generalised additive models) to unify new and existing extensions, and conveniently implement the models in the popular, open-source mgcv R package. We explore non-linear patterns of movement and habitat selection, and use the equivalence between penalised smoothing splines and random effects to implement individual-level and spatial random effects. This framework can also be used to fit varying-coefficient models to account for temporally or spatially-heterogeneous patterns of selection (e.g., resulting from behavioural variation), or any other non-linear interactions between drivers of the animal’s movement decisions. We provide the necessary technical details to understand several key special cases of smooths and their implementation in mgcv , showcase the ecological relevance using two illustrative examples, and provide R code (available at https://github.com/NJKlappstein/smoothSSF ) to facilitate the adoption of these methods. This paper is a broad overview of how smooth effects can be applied to increase the flexibility and biological realism of step selection analysis.
- 23 June 2023
- Published by Springer Nature in Marine Biotechnology
- Vol. 25 (4) , 580-587
SNP Panel and Genomic Sex Identification in Atlantic Halibut (Hippoglossus hippoglossus)
The ability to identify sex is necessary in population biology for a proper understanding of the dynamics of a population. In Atlantic halibut, phenotypic sex identification is not possible due to the lack of significant external morphological differences. We developed an Illumina SNP panel for Atlantic halibut with 4000 SNPs spread evenly throughout the genome with a minor allele frequency MAF ≥ 0.4, except for N = 249 SNPs located in a sex-determining region on chromosome 12, N = 176 of these SNPs were selected to genetically identify male and female individuals using a DAPC analysis. The genomic identification of sex allows for non-lethal sex determination and validation of sex identification in the field. The SNP panel is a new genomic resource for Atlantic halibut that will make it possible to generate the genotypic data for the large number of individuals needed to estimate population abundance using genomics and the Close Kin Mark Recapture (CKMR) approach, an emerging component of fisheries management and stock monitoring.
- 23 June 2023
- Published by Springer Nature in Marine Biotechnology
- Vol. 25 (4) , 580-587
SNP Panel and Genomic Sex Identification in Atlantic Halibut (Hippoglossus hippoglossus)
The ability to identify sex is necessary in population biology for a proper understanding of the dynamics of a population. In Atlantic halibut, phenotypic sex identification is not possible due to the lack of significant external morphological differences. We developed an Illumina SNP panel for Atlantic halibut with 4000 SNPs spread evenly throughout the genome with a minor allele frequency MAF ≥ 0.4, except for N = 249 SNPs located in a sex-determining region on chromosome 12, N = 176 of these SNPs were selected to genetically identify male and female individuals using a DAPC analysis. The genomic identification of sex allows for non-lethal sex determination and validation of sex identification in the field. The SNP panel is a new genomic resource for Atlantic halibut that will make it possible to generate the genotypic data for the large number of individuals needed to estimate population abundance using genomics and the Close Kin Mark Recapture (CKMR) approach, an emerging component of fisheries management and stock monitoring.
- Open Access
- 9 November 2022
- Published by Wiley in Canadian Journal of Statistics
- Vol. 50 (4) , 1355-1386
Canadian contributions to environmetrics
- 9 November 2022
- Published by Wiley in Canadian Journal of Statistics
- Vol. 50 (4) , 1355-1386
Open AccessCanadian contributions to environmetrics