Hierarchical occupancy models as solutions to challenges in biodiversity assessment

doi:10.17520/biods.2025386

Abstract

Abstract:

Background: Amid the accelerating global biodiversity crisis driven by human activities, precise monitoring and assessment of species distributions and population dynamics have become urgent priorities for conservation. Traditional survey methods often suffer from imperfect detection, leading to biased estimates of occupancy and hindering effective management decisions. The advent of big data offers opportunities for integrating diverse sources, yet challenges in handling heterogeneity and observational biases persist. Hierarchical occupancy models, by explicitly separating ecological processes (true occupancy) from observation processes (detection probability), provide a robust statistical framework to obtain unbiased inferences and have emerged as a powerful tool in biodiversity monitoring.

Main Content: This paper reviews the theoretical foundations of hierarchical occupancy models, including the classic single-season framework and key extensions such as multi-season (dynamic) models for quantifying colonization, extinction, and temporal trends, as well as multispecies (community) models that harness interspecific correlations to enhance inferential precision. We highlight their core advantages: correcting for false negatives through explicit detection probability estimation, flexibly integrating heterogeneous multi-source data, and generating interpretable, auditable ecological indicators for biodiversity assessment. However, practical applications face several challenges, including data quality and heterogeneity issues, violations of key assumptions (e.g., independence of observations, population closure within seasons, absence of false positives), potential constraints on the biological interpretability of parameters, high computational demands for complex models and large datasets, and difficulties in communicating results to non-specialists and policymakers. Corresponding mitigation strategies are discussed, such as standardized data preprocessing, rigorous assumption validation, interdisciplinary collaboration, algorithmic optimization, and enhanced science-policy translation.

Conclusion: Hierarchical occupancy models significantly advance the scientific rigor and reliability of biodiversity monitoring and evaluation by addressing imperfect detection and enabling integrative analyses. Moving forward, continued methodological innovations, fusion with emerging data types and technologies, deeper cross-disciplinary integration, and efforts toward standardization and broader application will further strengthen their role in supporting evidence-based conservation in the face of ongoing global change.

Key words: hierarchical occupancy model, biodiversity monitoring, imperfect detection, population dynamics, systematic conservation planning, multi-source data integration

Chunying Wu, Viorel D. Popescu, Yinqiu Ji. Hierarchical occupancy models as solutions to challenges in biodiversity assessment[J]. Biodiv Sci, 2026, 34(1): 25386.

Add to citation manager EndNote|Ris|BibTeX

URL: https://www.biodiversity-science.net/EN/10.17520/biods.2025386

https://www.biodiversity-science.net/EN/Y2026/V34/I1/25386

Figures/Tables 6

Fig. 1 Hierarchical occupancy model (HOM) addressing biodiversity monitoring challenges in the context of the Sixth Mass Extinction

Fig. 2 Simulation of hierarchical occupancy model (HOM) performance in correcting for imperfect detection. (a) Comparison of occupancy probability (ψ) estimates among different methods (sample size n = 100). The naive estimate systematically underestimates occupancy by ignoring imperfect detection, whereas the HOM estimate is closer to the true value. (b) Distribution of detection frequency at sites where the species is truly present. The bar at “0 detections” indicated by the red arrow represents “false absences”, where the species exists at the site but remains undetected due to a detection probability P < 1. This phenomenon is the primary source of bias in occupancy estimation.

Fig. 3 Dual-level statistical framework of the hierarchical occupancy model. This figure illustrates the dual-level statistical framework of the hierarchical occupancy model, which centers on correcting for imperfect detection in species surveys by decoupling the ecological and observation processes. The ecological process on the left describes the true occupancy state (latent variable zi) driven by environmental factors; the occupancy probability ψ is influenced by environmental covariates—such as habitat type (e.g., forest cover, vegetation structure), abiotic factors, and human disturbance—and is modeled via a logit link function. The observation process on the right describes the detection probability P during field surveys, which is affected by observational covariates including survey effort (duration, intensity, and number of replicates), weather conditions, and observer experience, also linked through a logit function. Given that the observation y is strictly dependent on the true occupancy state z (meaning a species can only be detected if it is truly present), the model integrates these processes by marginalizing over the latent variable zi to construct a joint likelihood function. This approach mathematically accounts for false absences, thereby achieving an unbiased estimation of the true species distribution.

Fig. 4 Schematic diagram of multi-season occupancy dynamics and state transition trends. This diagram illustrates the quarterly dynamics (where Q1-Q4 represent the first to fourth quarters of each year, respectively) of a species based on simulated data between 2015 and 2019: The occupancy rate (ψ, blue solid line) is generally maintained between 0.6-0.8, indicating a high persistent probability of presence in the study area. The colonization probability (γ, orange dotted line) is stable at 0.3-0.4 with a slight upward trend, reflecting a relatively strong capacity for range expansion. The extinction probability (ε, green dashed line) is maintained at 0.1-0.2, suggesting potential habitat disturbance or environmental pressure. This figure is created using simulated data and serves to illustrate the application of the multi-season hierarchical occupancy model (MSOM) in revealing species dynamics. Please see Appendix 2 for the R code.

Table 1 Core assumptions of hierarchical occupancy models

类别 Category	核心假设 Core assumption	描述 Description
生态过程 Ecological process	闭合假设 Closure assumption	在调查期间, 站点占有状态无变化(单季节模型); 动态模型放松此假设(MacKenzie et al., 2003) The occupancy status of a site remains constant during the survey period (single-season model); dynamic models relax this assumption (MacKenzie et al., 2003)
生态过程 Ecological process	独立站点 Site independence	站点间占有率条件独立, 或通过随机效应建模空间相关 Occupancy is conditionally independent among sites, or spatial correlation is accounted for through random effects
观测过程 Observation process	无假阳性 No false positives	检测到即真实存在; eDNA等场景需扩展模型处理假阳性(Guillera-Arroita et al., 2017) The species is truly present if detected; scenarios such as eDNA require model extensions to handle false positives (Guillera-Arroita et al., 2017)
观测过程 Observation process	检测独立 Independent detections	重复调查间检测事件独立 Detection events are independent across replicate surveys
参数估计 Parameter estimation	局部识别 Local identifiability	重复调查足够多以分离ψ和P (通常j ≥ 3) A sufficient number of replicate surveys are required to distinguish ψ from P (typically j ≥ 3)

Table 2 Framework for model selection and validation in hierarchical occupancy models

类别 Category	核心内容 Core item	描述 Description
模型选择 Model selection	AICc/WAIC/LOO-CV	用于比较模型拟合与复杂度; AICc适用于小样本(Burnham & Anderson, 2002), WAIC和LOO-CV适用于贝叶斯模型(Gelman et al., 2014; Vehtari et al., 2017) Used to compare model fit and complexity; AICc is suitable for small sample sizes (Burnham & Anderson, 2002), while WAIC and LOO-CV are appropriate for Bayesian models (Gelman et al., 2014; Vehtari et al., 2017)
模型选择 Model selection	后验预测检查 Posterior predictive check (PPC)	贝叶斯框架下评估模型拟合(Gelman et al., 2014) Assessing model fit within a Bayesian framework (Gelman et al., 2014)
模型验证 Model selection	交叉验证 Cross-validation (CV)	使用k折交叉验证评估预测性能 Evaluating predictive performance using k-fold cross-validation (k-fold CV)
残差诊断 Residual diagnostics	DHARMa残差 DHARMa residuals	模拟残差检查拟合优度, 适用于层级模型(Hartig, 2024) Using simulated residuals to check goodness-of-fit, specifically designed for hierarchical models (Hartig, 2024)

References 56

[1]	Beger M, Grantham HS, Pressey RL, Wilson KA, Peterson EL, Dorfman D, Mumby PJ, Possingham HP (2010) Conservation planning in a changing world: Multispecies approaches to prioritizing the global protection of biodiversity. BioScience, 60, 52-62.
[2]	Belmont J, Martino S, Illian J, Rue H (2024) Spatio-temporal occupancy models with INLA. Methods in Ecology and Evolution, 15, 2087-2100. DOI URL
[3]	Brambilla M, Caprio E, Assandri G, Scridel D, Bassi E, Bionda R, Celada C, Falco R, Bogliani G, Pedrini P, Rolando A, Chamberlain D (2017) A spatially explicit definition of conservation priorities according to population resistance and resilience, species importance and level of threat in a changing climate. Diversity and Distributions, 23, 727-738. DOI URL
[4]	Burnham KP, Anderson DR (2002) Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach, 2nd edn. Springer, New York.
[5]	Ceballos G, Ehrlich PR, Barnosky AD, García A, Pringle RM, Palmer TM (2015) Accelerated modern human-induced species losses: Entering the sixth mass extinction. Science Advances, 1, e1400253.
[6]	Chen IC, Hill JK, Ohlemüller R, Roy DB, Thomas CD (2011) Rapid range shifts of species associated with high levels of climate warming. Science, 333, 1024-1026.
[7]	Crist TO, Veech JA (2006) Additive partitioning of rarefaction curves and species-area relationships: Unifying α-, β- and γ-diversity with sample size and habitat area. Ecology Letters, 9, 923-932. DOI URL
[8]	Devarajan K, Morelli TL, Tenan S (2020) Multi-species occupancy models: Review, roadmap, and recommendations. Ecography, 43, 1612-1624. DOI URL
[9]	Dorazio RM, Gotelli NJ, Ellison AM (2011) Modern methods of estimating biodiversity from presence-absence surveys. In: BiodiversityLoss in a Changing Planet (ed. Grillo O), pp.71-102. InTech, Rijeka.
[10]	Dorazio RM, Royle JA, Söderström B, Glimskär A (2006) Estimating species richness and accumulation by modeling species occurrence and detectability. Ecology, 87, 842-854. PMID
[11]	Doser JW, Leuenberger W, Sillett TS, Hallworth MT, Zipkin EF (2022) Integrated community occupancy models: A framework to assess occurrence and biodiversity dynamics using multiple data sources. Methods in Ecology and Evolution, 13, 919-932. DOI URL
[12]	Ficetola GF, Pansu J, Bonin A, Coissac E, Giguet-Covex C, De Barba M, Gielly L, Lopes CM, Boyer F, Pompanon F, Rayé G, Taberlet P (2015) Replication levels, false presences and the estimation of the presence/absence from eDNA metabarcoding data. Molecular Ecology Resources, 15, 543-556. DOI URL
[13]	Ficetola GF, Taberlet P (2023) Towards exhaustive community ecology via DNA metabarcoding. Molecular Ecology, 32, 6320-6329. DOI PMID
[14]	Gelman A, Hill J (2007) Data Analysis Using Regression and Multilevel/Hierarchical Models. Cambridge University Press, Cambridge.
[15]	Gelman A, Hwang J, Vehtari A (2014) Understanding predictive information criteria for Bayesian models. Statistics and Computing, 24, 997-1016.
[16]	Gronau QF, Sarafoglou A, Matzke D, Ly A, Boehm U, Marsman M, Leslie DS, Forster JJ, Wagenmakers EJ, Steingroever H (2017) A tutorial on bridge sampling. Journal of Mathematical Psychology, 81, 80-97. DOI PMID
[17]	Guillera-Arroita G, Lahoz-Monfort JJ, van Rooyen AR, Weeks AR, Tingley R (2017) Dealing with false-positive and false-negative errors about species occurrence at multiple levels. Methods in Ecology and Evolution, 8, 1081-1091. DOI URL
[18]	Hamer AJ, Schloegel LM, Rossauer B, Anyomi K, Stockwell MP, Mahony MJ, Clulow S, Banks SC (2021) Multi-species occupancy models reveal community assembly and wetland restoration in amphibians. Journal of Applied Ecology, 58, 607-616.
[19]	Hansen LM, Pfeifer M (2019) Machine learning in remote sensing: A comprehensive review. ISPRS Journal of Photogrammetry and Remote Sensing, 152, 166-177. DOI URL
[20]	Hartig F (2024) DHARMa: Residual Diagnostics for Hierarchical (Multi-level/Mixed) Regression Models. R Package Version 0.4.7. https://CRAN.R-project.org/package=DHARMa. (accessed on 2025-09-02)
[21]	Hepler SA, Erhardt RJ (2021) A spatiotemporal model for multivariate occupancy data. Environmetrics, 32, e2657.
[22]	Hepler SA, Yang BQ (2024) Occupancy models with autocorrelated detection heterogeneity. Environmental and Ecological Statistics, 31, 777-800.
[23]	IPBES (2019) Global Assessment Report on Biodiversity and Ecosystem Services. IPBES Secretariat, Bonn, Germany.
[24]	Ji YQ, Baker CCM, Popescu VD, Wang JX, Wu CY, Wang ZY, Li YH, Wang L, Hua CL, Yang ZX, Yang CY, Xu CCY, Diana A, Wen QZ, Pierce NE, Yu DW (2022) Measuring protected-area effectiveness using vertebrate distributions from leech iDNA. Nature Communications, 13, 1555. DOI PMID
[25]	Ji YQ, Diana A, Li XY, Matechou E, Griffin JE, Liu SW, Luo MJ, Wu CY, Bai R, Yao CY, Yin TT, Dong F, Wu F, Wang K, Yu ZB, Chen XY, Jiang XL, Che J, Yu DW, Popescu VD (2025) High quality, granular, timely, trustworthy and efficient vertebrate species distribution data across a 30,000 km² protected area complex. Ecology Letters, 28, e70302.
[26]	Kelleher S, Guillera-Arroita G, Elith J, Briscoe NJ (2025) Twenty years of dynamic occupancy models: A review of applications and look to the future. Ecography, 48, e07757.
[27]	Kéry M, Royle JA (2015) Applied Hierarchical Modeling in Ecology: Analysis of Distribution, Abundance and Species Richness in R and BUGS (Vol. 1): Prelude and Static Models. Academic Press, New York.
[28]	Kéry M, Royle JA (2020) Applied Hierarchical Modeling in Ecology: Analysis of Distribution, Abundance and Species Richness in R and BUGS (Vol. 2): Dynamic and Advanced Models. Academic Press, New York.
[29]	Kéry M, Royle JA, Plattner M, Dorazio RM (2009) Species richness and occupancy estimation in communities subject to temporary emigration. Ecology, 90, 1279-1290. PMID
[30]	Kleiven EF, Barraquand F, Gimenez O, Henden JA, Ims RA, Soininen EM, Yoccoz NG (2023) A dynamic occupancy model for interacting species with two spatial scales. Journal of Agricultural, Biological and Environmental Statistics, 28, 466-482. DOI
[31]	Lipsey MK, Naugle DE, Nowak J, Lukacs PM (2017) Extending utility of hierarchical models to multi-scale habitat selection. Diversity and Distributions, 23, 783-793. DOI URL
[32]	MacKenzie DI, Nichols JD, Hines JE, Knutson MG, Franklin AB (2003) Estimating site occupancy, colonization, and local extinction when a species is detected imperfectly. Ecology, 84, 2200-2207. DOI URL
[33]	MacKenzie DI, Nichols JD, Lachman GB, Droege S, Royle JA, Langtimm CA (2002) Estimating site occupancy rates when detection probabilities are less than one. Ecology, 83, 2248-2255.
[34]	MacKenzie DI, Nichols JD, Royle JA, Pollock KH, Bailey LL, Hines JE (2017) Occupancy Estimation and Modeling: Inferring Patterns and Dynamics of Species Occurrence, 2nd edn. Academic Press, London.
[35]	MacKenzie DI, Nichols JD, Seamans ME, Gutiérrez RJ (2009) Modeling species occurrence dynamics with multiple states and imperfect detection. Ecology, 90, 823-835. PMID
[36]	MacKenzie DI, Royle JA (2005) Designing occupancy studies: General advice and allocating survey effort. Journal of Applied Ecology, 42, 1105-1114. DOI URL
[37]	Margules CR, Pressey RL (2000) Systematic conservation planning. Nature, 405, 243-253. DOI
[38]	Miller DA, Nichols JD, McClintock BT, Grant EHC, Bailey LL, Weir LA (2011) Improving occupancy estimation when two types of observational error occur: Non-detection and species misidentification. Ecology, 92, 1422-1428. PMID
[39]	Moilanen E, Wilson KA, Possingham HP (2009) Spatial Conservation Prioritization. Oxford University Press, Oxford.
[40]	Moll RJ, Kilshaw K, Montgomery RA, Abade L, Campbell RD, Harrington LA, Millspaugh JJ, Birks JDS, MacDonald DW (2016) Clarifying habitat niche width using broad-scale, hierarchical occupancy models: A case study with a recovering mesocarnivore. Journal of Zoology, 300, 177-185. DOI URL
[41]	Nichols JD, Williams BK (2006) Monitoring for conservation. Trends in Ecology & Evolution, 21, 668-673. DOI URL
[42]	Popescu VD, de Valpine P, Butsie DC (2016) Biological realism of wildlife population estimates in data-poor systems. Ecological Applications, 26, 1429-1444.
[43]	Rahman H, Moore JF, Connor T, Amin R, Evans MJ, Steinmetz R, Royle JA, Belant JL (2021) Assessing mammalian diversity in northeastern Bangladesh using multi-species occupancy models. Biodiversity and Conservation, 30, 15-34. DOI
[44]	Rota CT, Ferreira MAR, Kays RW, Forrester TD, Kalies EL, McShea WJ, Parsons AW, Millspaugh JJ (2016) A multispecies occupancy model for two or more interacting species. Methods in Ecology and Evolution, 7, 1164-1173. DOI URL
[45]	Royle JA, Dorazio RM (2008) Hierarchical Modeling and Inference in Ecology:The Analysis of Data from Populations, Metapopulations and Communities. Academic Press, San Diego.
[46]	Rue H, Martino S, Chopin N (2009) Approximate Bayesian inference for latent Gaussian models by using integrated nested Laplace approximations. Journal of the Royal Statistical Society Series B: Statistical Methodology, 71, 319-392. DOI URL
[47]	Saadi I, Farooq B, Mustafa A, Teller J, Cools M (2018) An efficient hierarchical model for multi-source information fusion. Expert Systems with Applications, 110, 352-362. DOI URL
[48]	Smith AM, Goldberg CS (2020) Occupancy in dynamic systems: Multi-scale and false-positive environmental DNA monitoring. Ecology, 101, e02945.
[49]	Sollmann R, Eaton MJ, Link WA, Mulondo P, Ayebare S, Prinsloo S, Plumptre AJ, Johnson DS (2021) A Bayesian Dirichlet process community occupancy model to estimate community structure and species similarity. Ecological Applications, 31, e02249.
[50]	Thuiller W, Guéguen M, Renaud J, Karger DN, Zimmermann NE (2019) Climate change and species distribution: From theory to practice. Global Change Biology, 25, 1126-1141.
[51]	Tittensor DP, Walpole M, Hill SLL, Boyce DG, Britten GL, Burgess ND, Butchart SHM, Leadley PW, Regan EC, Alkemade R, Baumung R, Bellard C, Bouwman L, Bowles-Newark NJ, Chenery AM, Cheung WWL, Christensen V, Cooper HD, Crowther AR, Dixon MJR, Galli A, Gaveau V, Gregory RD, Gutierrez NL, Hirsch TL, Höft R, Januchowski-Hartley SR, Karmann M, Krug CB, Leverington FJ, Loh J, Lojenga RK, Malsch K, Marques A, Morgan DHW, Mumby PJ, Newbold T, Noonan-Mooney K, Pagad SN, Parks BC, Pereira HM, Robertson T, Rondinini C, Santini L, Scharlemann JPW, Schindler S, Sumaila UR, Teh LSL, van Kolck J, Visconti P, Ye YM (2014) A mid-term analysis of progress toward international biodiversity targets. Science, 346, 241-244. DOI PMID
[52]	Tobler MW, Zúñiga Hartley A, Carrillo-Percastegui SE, Powell GVN (2015) Spatiotemporal hierarchical modelling of species richness and occupancy using camera trap data. Journal of Applied Ecology, 52, 413-421. DOI URL
[53]	Vehtari A, Gelman A, Gabry J (2017) Practical Bayesian model evaluation using leave-one-out cross-validation and WAIC. Statistics and Computing, 27, 1413-1432.
[54]	Wilson KA, Underwood EC, Morrison SA, Klausmeyer KR, Murdoch WW, Reyers B, Wardell-Johnson G, Marquet PA, Albert AS, Simons AJ (2007) Prioritizing global conservation efforts. Nature, 446, 109-112.
[55]	Zheng J (2023) Impact of transportation infrastructure on amphibians. Theoretical and Natural Science, 6, 426-430. DOI URL
[56]	Zipkin EF, DeWan A, Andrew Royle J (2009) Impacts of forest fragmentation on species richness: A hierarchical approach to community modelling. Journal of Applied Ecology, 46, 815-822. DOI URL