Lessons to be learned by comparing integrated fisheries stock assessment models (SAMs) with integrated population models (IPMs)

Fisheries Research, Journal Year: 2024, Volume and Issue: 272, P. 106925 - 106925

Published: Jan. 5, 2024

Integrated fisheries stock assessment models (SAMs) and integrated population (IPMs) are used in biological ecological systems to estimate abundance demographic rates. The approaches fundamentally very similar, but historically have been considered as separate endeavors, resulting a loss of shared vision, practice progress. We review the two identify similarities differences, with view identifying key lessons that would benefit more generally overarching topic ecology. present case study for each SAM (snapper from west coast New Zealand) IPM (woodchat shrikes Germany) highlight differences similarities. between SAMs IPMs appear be objectives parameter estimates required meet these objectives, size spatial scale populations, differing availability various types data. In addition, up now, typical applied aquatic habitats, while most stem terrestrial habitats. aim assess level sustainable exploitation fish so absolute or biomass must estimated, although some only relative trends. Relative is often sufficient understand dynamics inform conservation actions, which main objective IPMs. small populations concern, where uncertainty can important, conveniently implemented using Bayesian approaches. typically at moderate scales (1 104 km2), possibility collecting detailed longitudinal individual data, whereas large, economically valuable stocks large (104 106 km2) limited There sense data- (or information-) hungry than an because its goal abundance, data rates difficult obtain (often marine) applied. therefore require 'tuning' assumptions IPMs, 'data speak themselves', consequently techniques such weighting model evaluation nuanced being fit disaggregated quantify variation allow richer inference on processes. attempts example by unconditional capture-recapture

Language: Английский

---

Realising the promise of large data and complex models

Methods in Ecology and Evolution, Journal Year: 2023, Volume and Issue: 14(1), P. 4 - 11

Published: Jan. 1, 2023

In an era of rapid change, ecologists are increasingly asked to provide answers big, urgent questions global concern (Solé & Levin, 2022; Sutherland et al., 2013; Yates 2018). Concurrently, technological advances allow ecological data be collected at high resolutions (e.g. temporal and/or spatial scales), leading both new types and larger datasets becoming available (Farley These the opportunity investigate new, even previously unanswerable, questions, including those concerning animal movements (Nathan 2022) addressing conservation sustainability issues (Runting 2022). Increasingly, realistic models need developed fitted these (Fer 2018), pushing boundaries type intricacy that can explored (Niu 2020). However, big lead troubles across multiple aspects, from storing processing fitting complex interpreting output. Close collaborations between ecologists, statisticians, mathematical modellers, computer scientists other disciplines offer exciting ways forward solve problems, mutually beneficial advancements. For example, aid in efficient storage extraction data, development algorithms; statisticians help guide analysis via computational algorithms propagating or quantifying uncertainties throughout process; mathematicians ensure constructed most suitable fashion for specific demonstrate properties (such as territorial ranges population predictions); on biological characteristics systems studied interpretation corresponding results, thus informing future influencing policy decisions. The answer important is unprecedented, with declines biodiversity ecosystem services which will impact our ability meet Sustainable Development Goals (Reyers Selig, 2020), it through interdisciplinary biggest steps made. Data challenges arise full analytic pipeline, visualising developing ecologically relevant interpretable fit adapting associated efficiently obtaining meaningful interpretations practice, there often many trade-offs different aspects due during pipeline. within initial decisions may made regarding cleaning remove recorded errors) summarised form processed report scale). This itself challenging uncertainty process, potential errors being introduced. typically model data. motion-sensor camera trap a trade-off level (i.e. advanced tools used uniquely identifying individuals via, e.g. machine learning techniques) incorporate amount preprocessed assuming no error matches; incorporating matching uncertainty; allowing marked unmarked individuals). Alternatively, require computationally intensive them not scale increase size. consideration simpler more easily fitted, reducing fine-detail extracted data; adaptations model-fitting process such using some approximate approach aims robust approximations used, but potentially could biased parameter estimates. Special Feature provides combination review papers scientific articles address one modern day analyses large Echoing facing discipline, we present natural statistical cycle, starting limitations (and packages) presentation outputs. We consider each themes identified turn relating (i) (ii) model-fitting; (iii) visualisation interpretation. also emphasise very closely interlinked although have coarse 'pigeonholes', overlapping challenges. Ecology, like environmental sciences branches biology, has entered into enormous possibilities better understanding state 'big' characteristics. 'Four Vs Framework' (see discussion Farley al. (2018) references therein) discuss four distinct aspects: (1) volume: quantity (2) velocity: time-varying (3) variety: relationships; (4) veracity: trustworthiness do occur isolation, intricate when analysing highlight just problems approaches 'V' authors this encountered discussed. Biologging sensor technologies been forefront creating volumes frequently range scales. Thus, biologging pioneering ecology relation rapidly transform ecology, particularly their application Williams A key limitation current systems, however, collecting ultra-fine sub-second movement behaviour over shorter periods time vs. longer-term space use Wild (2023) take advantage developments field Internet Things methods attaching electronic devices, connected network, everyday objects) overcome networking Wi-Fi solutions, combined smart embedded software, able orders magnitude improvement retrieval efficiency, systems. particular, detail solutions software architecture, on-board difficulties synchronisation transmission concept pros cons infrastructures. Advances technology led (perhaps less foreseen) forms gathering mechanisms gaining momentum, build-up quantities rise citizen (or community) science initiatives. resulting initiatives varied nature, involving collection protocols limited/reduced structure than compared traditional survey methods, arising opportunistic events. While designed surveys requires carefully markedly semi-structured projects, example without fixed by observers any degree observer knowledge. leads whole spectrum 4 'V's. commonality terms similar overcome, expanse techniques, vary. Johnston summarise overarching categories challenges: behaviour, including, bias, reporting differences, false-positive errors; structures, measures detectability procedures validation; models, only opportunities provided integration multispecies sources bias limitations; (iv) communication, motivated monitoring. veracity arises obvious ways, outside sphere 'in field', commonly considered reason querying wealth information contained databases. combine information, must species taxa. raises methodological challenges, to, dynamic names, discovery species, changing attributes, etc. As result, homonyms, synonyms accumulate while taxa general consensus accepted name taxonomic phylogenetic relationships reached so taxonomy resemble confusingly tangled bank. To issues, Grenié extensive tools, databases best practices harmonising taxon names studies. they categorise 'wild world' existing publicly resources, along axes breadth scope, strengths caveats database. addition, practical computation side, R packages harmonisation, and, perhaps rather fittingly, 'taxonomy' packages, classifying according functions. vast array last decade (Guisan 2017; Hooten Kery Royle, 2016; MacKenzie 2018; McCrea Morgan, 2015; Royle 2014; Schaub Kéry, 2021), limited critical disadvantages each. advancement reflect (as highlighted above), quality increased spatial/temporal resolution), emerging earth observation drone eDNA) techniques power). summary overviews advancing areas timely understand what can, importantly, cannot should not), done encompass all community landscape ecology. Interrogation modelling ideas motivates further account additional complexities example. briefly here Feature. Developing applied scientifically efficient. Such permit readily NIMBLE (de Valpine 2017), R-INLA Lindgren Rue (2015) inlabru (Bachl 2019) well focused MARK/RMARK (for capture–recapture models; Laake, 2013); momentuHMM hidden Markov [HMMs] McClintock Michelot, 2018) Distance distance sampling; Thomas 2010). Areas accessible witnessing substantial development, enhanced flexibility provided. Laxton Torney (2023), Newman discusses relative merits models. Barros step issue suggesting primary challenge, predictive transparently adapted following updated methodology. Their proposal PERFICT workflow framework aligned. Understanding relationship challenging, structuring required form. Two particular 'umbrella' extensively related HMMs state-space (SSMs). Both widely settings presence longitudinal (Auger-Methe 2021; 2021). One attraction applications directly separate out sampling processes. simplifies specification, permitting components independently. common distinction relates whether latent processes defined discrete-valued HMMs) continuous-valued (SSMs), note universally used. Specific where applied, include, far fisheries stock assessment (Aeberhard 2018); dynamics (Newman 2014); (Hooten Langrock 2012; Patterson 2017); capture–recapture-type (King, 2015). Glennie practical) SSMs, respectively. specifying assumptions valid defining HMM underlying increases complexity. Providing descriptions variety aware useful resource practitioners, describing pitfalls arise. growth aided algorithms, Markovian (Zucchini 2016). flexible (latent) addressed (2023). Importantly, contrast wide-range dependent specified model. describe accommodate dynamics, nonlinear non-Gaussian stochasticity. familiar/used community, likely however SSMs great aspect paper. SSMs) expense, complexity increases. With datasets, routinely bioacoustics studies 2023), standard break down practically applied. There hence necessity identify develop modifications improve efficiency scalability, new) successful examples, strategies were successful, efficiencies 2023) demonstrated King (2022), simplifications retain signal promising avenues going forward. scalability algorithms. off-the-shelf sufficient too limiting, described Wang applications. generally known analysis. numerous opportunities, risks building structurally insight popular distribution (SDMs) importance increasing based theory. showcase usefulness point approach, permits inclusion linked covariates (relating study species), maintaining roles effects component, interpretability insight. show that, driving distributions substantially performance ability. Furthermore, non-monotonic, highlighting checking It now possible line drawn complexity's sake because output exhibited data? cases, simple actually useful/informative? long-standing areas, (Murtaugh, 2007). Statistical continue represent generating processes—but always simplification reality—with aiming extract general, (given data). learning) prominence usage (Ho Goethals, Pichler Hartig, 2022), demonstrating good performance, lack parameters. results/output appropriate real intelligent visualisations, beyond wider policy-makers area species' modelling. Traditionally, establish correlation single environment occupies order gain habitat suitability, predict impacts change. growing interest go isolation include interactions (Kissling Pollock 2014) (Buckley 2010) predictability model, parameters become difficult. issue, Powell-Romero feature-based ensemble Through features communities, obtain outperform others, why case. SDMs, argue needs grounded greater since patterns component dimensional, visualisations outputs improved becomes challenging. Traditional dimension reduction considering pair-wise correlations, nuanced insights masked, biases (McInerny McInerny Krzywinski, data/model networks graphs structures. food web us foodwebs complexity; tend simplify therefore needed. Pawluczuk Iskrzyński propose foodweb network) structures combining heatmaps, interactive animated graphs. Van Moorter package ConScape (in Julia) allows users analyse visualise connectivity simply. Further attempting objects contain (non-independent) parts make up complete object skeletons individual bones). focus, method regularised principal compare shape variation multipart morphospaces. accompanying package, community. data) immense. fully capitalise achieving academic societal impact, multidisciplinary pipeline required. number contribute knowledge (though exhaustive list): Immersive interdisciplinarity community's research largest step-changes discipline. cross-fertilisation from, engineers (designing devices), (developing exploit designing strategies) (offering expertise automation) co-creation problems. collaboration connection theory; equally stage build confidence results biologically realistic. diverse sizeable amounts (Zipkin continues pace, necessarily lagged timescale (there exist collected). Again, outlook novel yet towards integrating recent years (Frost spanning (Isaac 2020) (Barraquand Gimenez, 2019). means think about indeed comparable—do differing affect performance? small structured unstructured Global Biodiversity Information Facility (GBIF), limit latter, context dependency former 2020)? phrase, attributed statistician George Box, apply reasoning idea (accessible software) does mean useful) philosophy 'should we' given dataset, ask necessary question Gain trump sophistication per se. role direction domain (Pichler prediction objective. simply blindly align analytical trends—it driver underpinning usage. 'black-box' nature constraints input, Considerable debate remains validity generalisability, conceptual simplicity, robustness transparency. efforts artificial intelligence power appropriately harnessed evolution. interpreted likelihood-based methods. prominent feature analyses. conducted Bachl 2019 de specialised (Van 2023). languages, Python problem hand. Clear guidance advantages resource, though difficult practice. communication solving inherent focus disseminating circle technical groups. moving code sharing, investing teaching activities resources. They conclude 'democratisation' emulate progress brought democratisation pressing time. scratch surface dealing advance understanding. rich set researchers, recognising pipelines providing produced cross-disciplinary reach its potential, understanding, firm basis informed decision-making. special arose discussions ICMS-funded meeting 'Addressing Challenges Modern Technological Advances', organised National Centre joint BES Quantitative Movement Ecology Interest Group Meeting Sheffield 2018. RM currently funded EPSRC grant EP/S020470/1. peer history article https://www.webofscience.com/api/gateway/wos/peer-review/10.1111/2041-210X.14050.

Language: Английский

---

Robust abundance estimation of harvested populations from low quality time series data: A red deer case study

Ecological Indicators, Journal Year: 2025, Volume and Issue: 173, P. 113398 - 113398

Published: April 1, 2025

Language: Английский

---

Population dynamic life history models of the birds and mammals of the world

Ecological Informatics, Journal Year: 2024, Volume and Issue: 80, P. 102492 - 102492

Published: Jan. 29, 2024

With life history traits determining the natural selection fitnesses of individuals and growth populations, estimates their variation are essential to advance evolutionary understanding ecological management during times global change. As data incomplete or missing for most species, I combine theory construct a meta model population dynamic (PDLH) in birds mammals. This generates PDLH models 11,187 species 4936 mammals, covering 29 per species. The inter-specific is used illustrate underlying mechanisms, explain diverse range trajectories by inclusion regulation. provides steps towards improved analyses freely accessible ready-to-use online simulations,

Language: Английский

---

Forecasting insect dynamics in a changing world

Current Opinion in Insect Science, Journal Year: 2023, Volume and Issue: 60, P. 101133 - 101133

Published: Oct. 17, 2023

Predicting how insects will respond to stressors through time is difficult because of the diversity insects, environments, and approaches used monitor model. Forecasting models take correlative/statistical, mechanistic models, integrated forms; in some cases, temporal processes can be inferred from spatial models. Because heterogeneity associated with broad community measurements, are often unable identify explanations. Many present efforts forecast insect dynamics restricted single-species which offer precise predictions but limited generalizability. Trait-based may a good compromise that limits masking ranges responses while still offering insight. Regardless modeling approach, data parameterize forecasting model should carefully evaluated for autocorrelation, minimum needs, sampling biases data. tested using near-term revised improve future forecasts.

Language: Английский

---

A crowded ocean: The need for demographic and movement data in seabird conservation

Ocean & Coastal Management, Journal Year: 2023, Volume and Issue: 244, P. 106833 - 106833

Published: Sept. 1, 2023

Language: Английский

---

First estimate of distribution, abundance, and risk of encounter with aquaculture vessels for the rare Chilean dolphin in the entire Northern Chilean Patagonia

Aquatic Conservation Marine and Freshwater Ecosystems, Journal Year: 2023, Volume and Issue: 33(12), P. 1535 - 1551

Published: Sept. 12, 2023

Abstract Assessing distribution and abundance patterns for rare species is challenging yet imperative, considering the extant, potentially hazardous, anthropogenic activities. In particular, poorly studied Chilean dolphin ( Cephalorhynchus eutropia ) exhibits an extremely patchy low densities, co‐occurring with intensive aquaculture industry in Southern Chile, among other The of dolphins were assessed entire Northern Patagonia. A hierarchical model was fitted to data from line transect surveys using distance sampling techniques environmental variables derived topographic features oceanographic models. second version this used a joint‐likelihood approach incorporate presence–pseudoabsence improving parameter estimation. Spatial predictions arising these models evaluate relative probability encountering vessels local, predominantly fleet. results show that drastically reduced uncertainty parameters controlling effect covariates total estimates. This estimated overall (median 2,225.8; 95% CI 1,340–3,867) region, indicates their preference shallow, sheltered bays inner channels, near river mouths, where salinity expected. highest probabilities dolphin–vessel interactions found on eastern coast Chiloe Big Island, coinciding largest number concessions area. Considering population expected be thousands, suitable habitat highly restricted, facing increasing impacts, some areas undergoing or planning major expansions development, provided here should considered management plans extant marine protected areas, evaluation current International Union Conservation Nature (IUCN) national conservation categories species.

Language: Английский

---

Modeling host–microbiome interactions to improve mechanistic understanding of aphid vectored plant pathogens

Frontiers in Ecology and Evolution, Journal Year: 2023, Volume and Issue: 11

Published: Sept. 27, 2023

Insect transmission of plant pathogens involves multi-layered interactions between vectors, viruses, host plants and environmental factors. Adding to the complexity vector–virus relationships are diverse microbial communities, which hypothesized influence pathogen transmission. Although interaction research has flourished, role played by microbes in vector competence disease epidemiology remains unclear many pathosystems. We therefore aimed develop a novel ecological modeling approach identify drivers complex vector–virus–microbiome interactions, particularly differences abundance symbionts within microbiomes probability virus acquisition. Our combines established molecular tools for profiling communities with underutilized Bayesian hierarchical data integration techniques. used globally relevant aphid–virus pathosystem custom vector–microbiome models that incorporate covariates (e.g., temperature, landcover) applied them individual extent factors drive changes then acquisition aphid. Specifically, we focus on aphid obligate symbiont ( Buchnera ) wide-spread facultative Serratia as proof concept two major species include single covariate (i.e., temperature). Overall, demonstrate how community-level microbiome can candidate variables associated competence. framework accommodate range different abundances, overcome spatial misalignment streams, is robust varying levels incidence. Results show relative strongly negatively S.avenae , but not R. padi . was competence, influenced spring temperatures. This work lays foundation developing broader predicting dynamics agroecosystems deploying microbiome-targeted pest management tactics.

Language: Английский

---

Integrated integral population models

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2023, Volume and Issue: unknown

Published: Oct. 28, 2023

Summary Data integration allows obtaining better descriptions and forecasting of a population’s behaviour by incorporating data at both the individual population levels. Structured models include matrix (MPMs), which structure through discrete state variable, integral (IPMs), use continuous variable. Two decades ago, integrated version MPMs appeared, but their corresponding for IPMs is still missing. Here, we propose model (IIPM). This takes up ideas behind existing used to describe forecast dynamics continuously structured populations: IPMs, data, inverse data. Particularly, emphasise construction fitting IIPM under Bayesian framework Soay sheep database compare generated these models. The constructed with had good performance (vital rates) (size structure) levels, because, as they are constrained fit sets produce balanced dynamics. In turn, IPM produced best estimates worst estimates, whilst estimates. objective should be correctly patterns. not using fail in this objective. An solves problem.

Language: Английский

---

Melded Integrated Population Models

Journal of Agricultural Biological and Environmental Statistics, Journal Year: 2024, Volume and Issue: unknown

Published: May 4, 2024

Language: Английский

---