pith. sign in

arxiv: 2606.26323 · v1 · pith:JCGMP7KZnew · submitted 2026-06-24 · 📊 stat.AP

False Positives, False Negatives, and the Detection-Only Problem: A Hierarchical Model for Species Occurrence with Observation Error

Kabiru Abubakari , Eleni Matechou , Marie C. Henniges , Ilia J. Leitch , Andrew R. Leitch , Silvia Liverani This is my paper

Pith reviewed 2026-06-26 00:48 UTC · model grok-4.3

classification 📊 stat.AP
keywords hierarchical modelspecies occurrencefalse positivefalse negativedetection-only dataidentifiabilityoccupancy modelobservation error
0
0 comments X

The pith

A hierarchical model for detection-only species data accounts for both false positives and false negatives via targeted priors.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper unifies occupancy models, species distribution models, and presence-only methods as special cases of one hierarchical observation process. It defines detection/non-detection data with T visits (DN-T) and detection-only data (DO), then introduces a new model for DO data that models both false positive and false negative errors for the first time. Identifiability is obtained by placing prior distributions on recording probabilities that encode the belief a species is more likely recorded where present than absent. This setup lets imperfect observation data support more reliable estimates of species occurrence for biodiversity monitoring.

Core claim

We argue that these are all special cases of a single hierarchical observation process. To make these connections explicit, we introduce a unified terminology centred on two data types: detection/non-detection data with T visits (DN-T) and detection-only data (DO), where DN-T with T>1 corresponds to traditional occupancy modelling, DN-1 to species distribution modelling, and DO to what the literature commonly, but we argue inaccurately, calls presence-only data. Within this framework, we study the identifiability of DO models and propose a novel hierarchical model for DO data that, for the first time, explicitly accounts for both false positive and false negative detection errors. Identifiab

What carries the argument

Hierarchical model for DO data with priors on recording probabilities that are higher conditional on true presence than on absence.

If this is right

  • Occupancy, species distribution, and presence-only approaches become interchangeable under one observation process.
  • Detection-only records can be analyzed while separating false positives from false negatives.
  • Species occurrence estimates from imperfect data become usable for conservation without requiring repeated visits.
  • The unified terminology allows direct comparison of results across different survey designs.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Data sets collected under different protocols could be pooled or compared more directly.
  • The strength of the identifying priors may need calibration against known cases to avoid over- or under-correction.
  • The same prior structure might extend to multi-species or spatially explicit versions of the model.

Load-bearing premise

The priors that encode higher recording probability where a species is present than absent are sufficient to achieve identifiability.

What would settle it

A simulation or field study with known true presence/absence locations where the posterior distributions for occupancy and detection parameters remain non-identifiable or produce biased estimates despite the specified priors.

Figures

Figures reproduced from arXiv: 2606.26323 by Andrew R. Leitch, Eleni Matechou, Ilia J. Leitch, Kabiru Abubakari, Marie C. Henniges, Silvia Liverani.

Figure 1
Figure 1. Figure 1: A hierarchical illustration of the species observation process underlying all occur￾rence data, regardless of survey design. ( ) represents latent states that are never directly observed: the occupancy state Zi ∈ {0, 1}, indicating whether the species occupies site si , and the site visit indicator Vi ∈ {0, 1}, indicating whether the site was visited, assumed indepen￾dent of Zi. ( ) represents detection-on… view at source ↗
Figure 2
Figure 2. Figure 2: For both the generalised linear model (GLM) and the detection-only model (DO [PITH_FULL_IMAGE:figures/full_fig_p018_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Posterior means and 95% credible intervals for the intercept and coefficients of [PITH_FULL_IMAGE:figures/full_fig_p020_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Predicted distribution maps for Dactylorhiza fuchsii (common spotted orchid) and Carex limosa (bog-sedge) from DO-M fitted to BSBI detection records at 10 km resolution. Predicted occurrence probabilities range from low (yellow) to high (purple). Orchis fuchsii is predicted to be widespread across England, Wales, and south of Scotland, whilst Carex limosa is predicted to be largely restricted to upland are… view at source ↗
Figure 5
Figure 5. Figure 5: Induced prior distributions under different specifications for [PITH_FULL_IMAGE:figures/full_fig_p035_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: (a) The intercept and coefficient estimates for DO-M and GLM as the background [PITH_FULL_IMAGE:figures/full_fig_p039_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: (a) Convergence of effect sizes and intercept estimates for [PITH_FULL_IMAGE:figures/full_fig_p042_7.png] view at source ↗
read the original abstract

Monitoring species occurrence is essential for understanding biodiversity change, informing conservation decisions, and assessing the impact of environmental pressures on ecosystems. Species occurrence data arise from different survey designs, and the statistical literature has developed distinct corresponding modelling approaches, namely occupancy models, species distribution models, and presence-only methods, whose fundamental connections have remained largely unrecognised. We argue that these are all special cases of a single hierarchical observation process. To make these connections explicit, we introduce a unified terminology centred on two data types: detection/non-detection data with T visits (DN-T) and detection-only data (DO), where DN-T with T>1 corresponds to traditional occupancy modelling, DN-1 to species distribution modelling, and DO to what the literature commonly, but we argue inaccurately, calls presence-only data. Within this framework, we study the identifiability of DO models and propose a novel hierarchical model for DO data that, for the first time, explicitly accounts for both false positive and false negative detection errors. Identifiability is achieved through prior distributions that express the natural belief that a species is more likely to be recorded where it is present than where it is absent. ...

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript unifies occupancy models, species distribution models, and presence-only methods as special cases of a single hierarchical observation process with two data types: detection/non-detection with T visits (DN-T) and detection-only (DO). It proposes a novel hierarchical model for DO data that explicitly accounts for both false-positive and false-negative errors and claims that identifiability is achieved for the first time by placing priors that encode the belief P(record|present) > P(record|absent).

Significance. If the identifiability claim holds with data-driven recovery rather than prior domination, the unification would connect previously separate literatures and enable more reliable use of common DO data for biodiversity monitoring and conservation. The framework itself is a useful conceptual contribution even if the DO model requires further validation.

major comments (2)
  1. [Abstract] Abstract (and the central claim): the statement that the chosen priors achieve identifiability 'for the first time' is not supported by either an analytic proof that the posterior separates the occurrence probability, true-positive rate, and false-positive rate globally, or by simulation results demonstrating parameter recovery at realistic sample sizes. The DO likelihood depends only on the product of occurrence and detection probabilities minus false-positive contributions, leaving the surface flat along the relevant manifold; priors can regularize but do not automatically supply the missing information.
  2. [Model description / Results] The manuscript does not appear to contain simulation experiments or analytic derivations that test whether posterior means and credible intervals recover known true values when data are generated from the model; without such evidence the practical utility of the DO model remains unestablished.
minor comments (2)
  1. [Model] Clarify the exact functional form of the hierarchical likelihood for DO data (including how false positives enter) and state whether any auxiliary assumptions beyond the priors are required.
  2. [Introduction] The unified terminology (DN-T, DN-1, DO) is helpful but should be accompanied by an explicit mapping table showing how each literature's standard model emerges as a special case.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed comments, which highlight important aspects of our work on the unified hierarchical framework for species occurrence data. We address each major comment below.

read point-by-point responses

  1. Referee: [Abstract] Abstract (and the central claim): the statement that the chosen priors achieve identifiability 'for the first time' is not supported by either an analytic proof that the posterior separates the occurrence probability, true-positive rate, and false-positive rate globally, or by simulation results demonstrating parameter recovery at realistic sample sizes. The DO likelihood depends only on the product of occurrence and detection probabilities minus false-positive contributions, leaving the surface flat along the relevant manifold; priors can regularize but do not automatically supply the missing information.

    Authors: We agree that the manuscript does not contain an analytic proof of global identifiability or simulation results demonstrating parameter recovery. The proposed identifiability relies on the informative priors that encode the belief P(record|present) > P(record|absent), which we argue separates the parameters in the posterior by breaking the flat manifold in the likelihood. However, we acknowledge that this requires empirical support to confirm data-driven recovery rather than prior domination. We will revise the abstract to remove or qualify the 'for the first time' phrasing and add simulation studies in the revised manuscript to demonstrate parameter recovery at realistic sample sizes. revision: yes

  2. Referee: [Model description / Results] The manuscript does not appear to contain simulation experiments or analytic derivations that test whether posterior means and credible intervals recover known true values when data are generated from the model; without such evidence the practical utility of the DO model remains unestablished.

    Authors: We acknowledge the absence of simulation experiments or analytic derivations testing parameter recovery in the current manuscript. To establish the practical utility of the DO model, we will conduct and include simulation studies that generate data from the model and evaluate whether posterior means and credible intervals recover the known true values under various scenarios. These results will be added to the revised version. revision: yes

Circularity Check

0 steps flagged

No circularity: identifiability claim rests on explicit prior construction rather than reduction to inputs

full rationale

The paper's central step is the introduction of a hierarchical model for DO data whose identifiability is asserted to follow from the choice of priors that encode P(record|present) > P(record|absent). This is a modeling assumption, not a derivation that reduces by construction to fitted parameters or prior self-citations. No equations are shown that equate a 'prediction' to a fitted quantity, and the abstract supplies no self-citation chain that bears the load of the uniqueness claim. The derivation chain therefore remains self-contained against external benchmarks; the Bayesian identifiability statement is a direct consequence of the stated prior family rather than a renaming or smuggling of an earlier result.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

Based on the abstract only, the central claim rests on the domain assumption that existing modeling approaches are special cases of one hierarchical process and the ad hoc choice of priors for identifiability. No free parameters or invented entities are explicitly described.

axioms (2)
  • domain assumption Occupancy models, species distribution models, and presence-only methods are all special cases of a single hierarchical observation process.
    Stated directly in the abstract as the fundamental unrecognized connections.
  • ad hoc to paper Prior distributions expressing that a species is more likely recorded where present than absent achieve identifiability for DO models.
    The abstract states that identifiability is achieved through these priors.

pith-pipeline@v0.9.1-grok · 5769 in / 1364 out tokens · 23125 ms · 2026-06-26T00:48:37.488606+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

94 extracted references

  1. [1]

    2006 , publisher=

    Uncertain judgements: eliciting experts' probabilities , author=. 2006 , publisher=

    2006

  2. [2]

    Journal of Computational and Graphical Statistics , year =

    Programming with models: writing statistical algorithms for general model structures with. Journal of Computational and Graphical Statistics , year =

  3. [3]

    Isaac, N. J. B. and van Strien, A. J. and August, T. A. and de Zeeuw, M. P. and Roy, D. B. , title =. Methods in Ecology and Evolution , year =

  4. [4]

    and Nichols, J

    Altwegg, R. and Nichols, J. D. , title =. Methods in Ecology and Evolution , year =

  5. [5]

    Doser, J. W. and Finley, A. O. and Banerjee, S. , title =. Global Ecology and Biogeography , year =

  6. [6]

    Ecological Modelling , volume=

    Incorporation of latitude-adjusted bioclimatic variables increases accuracy in species distribution models , author=. Ecological Modelling , volume=. 2022 , publisher=

    2022

  7. [7]

    Annals of Botany , volume=

    Evolution of genome size in Carex (Cyperaceae) in relation to chromosome number and genomic base composition , author=. Annals of Botany , volume=. 2013 , publisher=

    2013

  8. [8]

    2023 , school =

    Changes in the British and Irish flora: the role of genome size , author =. 2023 , school =

    2023

  9. [9]

    and Dirzo, Rodolfo , title =

    Ceballos, Gerardo and Ehrlich, Paul R. and Dirzo, Rodolfo , title =. Proceedings of the National Academy of Sciences , year =

  10. [10]

    IPBES secretariat , volume=

    Global assessment report on biodiversity and ecosystem services of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services , author=. IPBES secretariat , volume=. 2019 , publisher=

    2019

  11. [11]

    , title =

    Urban, Mark C. , title =. Science , year =

  12. [12]

    Miller, David A. W. and Pacifici, Krishna and Sanderlin, Jamie S. and Reich, Brian J. , title =. Methods in Ecology and Evolution , year =

  13. [13]

    , title =

    Neal, Radford M. , title =. Handbook of. 2011 , pages =

    2011

  14. [14]

    , title =

    Gelman, Andrew and Rubin, Donald B. , title =. Statistical Science , year =

  15. [15]

    Trends in ecology & evolution , volume=

    Data integration for large-scale models of species distributions , author=. Trends in ecology & evolution , volume=. 2020 , publisher=

    2020

  16. [16]

    Scientific Data , volume=

    A taxonomic, genetic and ecological data resource for the vascular plants of Britain and Ireland , author=. Scientific Data , volume=. 2022 , publisher=

    2022

  17. [17]

    Biometrical Journal , volume=

    Binary regression with misclassified response and covariate subject to measurement error: a Bayesian approach , author=. Biometrical Journal , volume=. 2008 , publisher=

    2008

  18. [18]

    Biodiversity informatics , volume=

    Presence-only and presence-absence data for comparing species distribution modeling methods , author=. Biodiversity informatics , volume=

  19. [19]

    arXiv preprint arXiv:1912.10981 , year=

    Missing data analysis and imputation via latent Gaussian Markov random fields , author=. arXiv preprint arXiv:1912.10981 , year=

    arXiv 1912

  20. [20]

    Ecography , volume=

    Novel methods improve prediction of species’ distributions from occurrence data , author=. Ecography , volume=. 2006 , publisher=

    2006

  21. [21]

    Ecological monographs , volume=

    Predictive performance of presence-only species distribution models: a benchmark study with reproducible code , author=. Ecological monographs , volume=. 2022 , publisher=

    2022

  22. [22]

    Methods in Ecology and Evolution , volume=

    Point process models for presence-only analysis , author=. Methods in Ecology and Evolution , volume=. 2015 , publisher=

    2015

  23. [23]

    Ecological applications , volume=

    Sample selection bias and presence-only distribution models: implications for background and pseudo-absence data , author=. Ecological applications , volume=. 2009 , publisher=

    2009

  24. [24]

    Journal of the Royal Statistical Society Series C: Applied Statistics , volume=

    Bayesian analysis of measurement error models using integrated nested Laplace approximations , author=. Journal of the Royal Statistical Society Series C: Applied Statistics , volume=. 2015 , publisher=

    2015

  25. [25]

    Biometrics , volume=

    Presence-only data and the EM algorithm , author=. Biometrics , volume=. 2009 , publisher=

    2009

  26. [26]

    Methods in Ecology and Evolution , volume=

    Local frequency as a key to interpreting species occurrence data when recording effort is not known , author=. Methods in Ecology and Evolution , volume=. 2012 , publisher=

    2012

  27. [27]

    Biodiversity Informatics , year =

    Presence-only and presence-absence data for comparing species distribution modeling methods , author =. Biodiversity Informatics , year =

  28. [28]

    Journal of applied ecology , volume=

    Modelling distribution and abundance with presence-only data , author=. Journal of applied ecology , volume=. 2006 , publisher=

    2006

  29. [29]

    Communications in Statistics-Theory and Methods , volume=

    Estimating logistic regression models when the dependent variable has no variance , author=. Communications in Statistics-Theory and Methods , volume=. 1992 , publisher=

    1992

  30. [30]

    Ecology , volume=

    Weighted distributions and estimation of resource selection probability functions , author=. Ecology , volume=. 2006 , publisher=

    2006

  31. [31]

    Journal of Econometrics , volume=

    Case-control studies with contaminated controls , author=. Journal of Econometrics , volume=. 1996 , publisher=

    1996

  32. [32]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Logistic methods for resource selection functions and presence-only species distribution models , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  33. [33]

    The annals of applied statistics , volume=

    Finite-sample equivalence in statistical models for presence-only data , author=. The annals of applied statistics , volume=

  34. [34]

    The Annals of Applied Statistics , volume=

    Analysis of presence-only data via exact Bayes, with model and effects identification , author=. The Annals of Applied Statistics , volume=. 2022 , publisher=

    2022

  35. [35]

    Proceedings of the twenty-first international conference on Machine learning , pages=

    A maximum entropy approach to species distribution modeling , author=. Proceedings of the twenty-first international conference on Machine learning , pages=

  36. [36]

    Ecography , volume=

    Modeling of species distributions with Maxent: new extensions and a comprehensive evaluation , author=. Ecography , volume=. 2008 , publisher=

    2008

  37. [37]

    Methods in ecology and evolution , volume=

    Selecting pseudo-absences for species distribution models: How, where and how many? , author=. Methods in ecology and evolution , volume=. 2012 , publisher=

    2012

  38. [38]

    Journal of the Royal Statistical Society Series C: Applied Statistics , volume=

    Point pattern modelling for degraded presence-only data over large regions , author=. Journal of the Royal Statistical Society Series C: Applied Statistics , volume=. 2011 , publisher=

    2011

  39. [39]

    Biometrics , volume=

    Predicting the geographic distribution of a species from presence-only data subject to detection errors , author=. Biometrics , volume=. 2012 , publisher=

    2012

  40. [40]

    Ecography , volume=

    BIOMOD--a platform for ensemble forecasting of species distributions , author=. Ecography , volume=. 2009 , publisher=

    2009

  41. [41]

    Journal of biogeography , volume=

    Making better Maxent models of species distributions: complexity, overfitting and evaluation , author=. Journal of biogeography , volume=. 2014 , publisher=

    2014

  42. [42]

    Ecological Modelling , volume=

    The influence of the number and distribution of background points in presence-background species distribution models , author=. Ecological Modelling , volume=. 2024 , publisher=

    2024

  43. [43]

    Diversity and Distributions , volume=

    A review of evidence about use and performance of species distribution modelling ensembles like BIOMOD , author=. Diversity and Distributions , volume=. 2019 , publisher=

    2019

  44. [44]

    Ecography , volume=

    Opening the black box: An open-source release of Maxent , author=. Ecography , volume=. 2017 , publisher=

    2017

  45. [45]

    Global Ecology and Biogeography , volume=

    Accounting for imperfect detection and survey bias in statistical analysis of presence-only data , author=. Global Ecology and Biogeography , volume=. 2014 , publisher=

    2014

  46. [46]

    Journal of Agricultural, Biological and Environmental Statistics , volume=

    Presence-only for marked point process under preferential sampling , author=. Journal of Agricultural, Biological and Environmental Statistics , volume=. 2024 , publisher=

    2024

  47. [47]

    Ecology , volume=

    On estimating probability of presence from use--availability or presence--background data , author=. Ecology , volume=. 2013 , publisher=

    2013

  48. [48]

    2002 , publisher=

    Resource selection by animals: statistical design and analysis for field studies , author=. 2002 , publisher=

    2002

  49. [49]

    Ecological modelling , volume=

    Maximum entropy modeling of species geographic distributions , author=. Ecological modelling , volume=. 2006 , publisher=

    2006

  50. [50]

    Methods in Ecology and Evolution , volume=

    Likelihood analysis of species occurrence probability from presence-only data for modelling species distributions , author=. Methods in Ecology and Evolution , volume=. 2012 , publisher=

    2012

  51. [51]

    Ecography , volume=

    Inference from presence-only data; the ongoing controversy , author=. Ecography , volume=. 2013 , publisher=

    2013

  52. [52]

    Journal of econometrics , volume=

    Misclassification of the dependent variable in a discrete-response setting , author=. Journal of econometrics , volume=. 1998 , publisher=

    1998

  53. [53]

    Journal of Econometrics , volume=

    Misclassification in binary choice models , author=. Journal of Econometrics , volume=. 2017 , publisher=

    2017

  54. [54]

    Ecology , volume=

    Generalized site occupancy models allowing for false positive and false negative errors , author=. Ecology , volume=. 2006 , publisher=

    2006

  55. [55]

    Journal of Applied Statistics , volume=

    Bayesian estimation of logistic regression with misclassified covariates and response , author=. Journal of Applied Statistics , volume=. 2018 , publisher=

    2018

  56. [56]

    Journal of the Royal Statistical Society Series C: Applied Statistics , volume=

    Modelling environmental DNA data; Bayesian variable selection accounting for false positive and false negative errors , author=. Journal of the Royal Statistical Society Series C: Applied Statistics , volume=. 2020 , publisher=

    2020

  57. [57]

    Ecography , volume=

    Climate-related range shifts--a global multidimensional synthesis and new research directions , author=. Ecography , volume=. 2015 , publisher=

    2015

  58. [58]

    Environmental Evidence , volume=

    Climate change and the global redistribution of biodiversity: substantial variation in empirical support for expected range shifts , author=. Environmental Evidence , volume=. 2023 , publisher=

    2023

  59. [59]

    Global Change Biology , volume=

    Species’ traits as predictors of range shifts under contemporary climate change: A review and meta-analysis , author=. Global Change Biology , volume=. 2017 , publisher=

    2017

  60. [60]

    science , volume=

    The biodiversity of species and their rates of extinction, distribution, and protection , author=. science , volume=. 2014 , publisher=

    2014

  61. [61]

    Conservation Biology , volume=

    A framework for evaluating the impact of the IUCN Red List of threatened species , author=. Conservation Biology , volume=. 2020 , publisher=

    2020

  62. [62]

    PloS one , volume=

    The effects of sampling bias and model complexity on the predictive performance of MaxEnt species distribution models , author=. PloS one , volume=. 2013 , publisher=

    2013

  63. [63]

    Methods Ecol Evol 6 (4): 424--438 , author=

    Bias correction in species distribution models: pooling survey and collection data for multiple species. Methods Ecol Evol 6 (4): 424--438 , author=

  64. [64]

    Methods in Ecology and Evolution , volume=

    Understanding co-occurrence by modelling species simultaneously with a Joint Species Distribution Model (JSDM) , author=. Methods in Ecology and Evolution , volume=. 2014 , publisher=

    2014

  65. [65]

    2020 , publisher=

    Joint species distribution modelling: With applications in R , author=. 2020 , publisher=

    2020

  66. [66]

    Ecology , volume=

    Estimating site occupancy rates when detection probabilities are less than one , author=. Ecology , volume=. 2002 , publisher=

    2002

  67. [67]

    2017 , publisher=

    Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence , author=. 2017 , publisher=

    2017

  68. [68]

    arXiv preprint arXiv:2501.18106 , year=

    Logistic regression models: practical induced prior specification , author=. arXiv preprint arXiv:2501.18106 , year=

    arXiv

  69. [69]

    2024 , url =

    Siegrist, Kyle , title =. 2024 , url =

    2024

  70. [70]

    Communications in Statistics-Simulation and Computation , volume=

    The need to conduct repeated classifications in a logistic regression model with misclassification in the dependent variable , author=. Communications in Statistics-Simulation and Computation , volume=. 2021 , publisher=

    2021

  71. [71]

    PeerJ , volume=

    Characterisation of false-positive observations in botanical surveys , author=. PeerJ , volume=. 2017 , publisher=

    2017

  72. [72]

    2025 , note =

    cmdstanr: R Interface to 'CmdStan' , author =. 2025 , note =

    2025

  73. [73]

    The Annals of Applied Statistics , volume=

    Models with observation error and temporary emigration for count data , author=. The Annals of Applied Statistics , volume=. 2024 , publisher=

    2024

  74. [74]

    Ecological Solutions and Evidence , volume=

    Predicting detection probabilities to estimate species' population sizes , author=. Ecological Solutions and Evidence , volume=. 2026 , publisher=

    2026

  75. [75]

    Ecological modelling , volume=

    Predicting species spatial distributions using presence-only data: a case study of native New Zealand ferns , author=. Ecological modelling , volume=. 2002 , publisher=

    2002

  76. [76]

    Journal of applied ecology , volume=

    An improved approach for predicting the distribution of rare and endangered species from occurrence and pseudo-absence data , author=. Journal of applied ecology , volume=. 2004 , publisher=

    2004

  77. [77]

    Ecography , volume=

    Effects of restricting environmental range of data to project current and future species distributions , author=. Ecography , volume=. 2004 , publisher=

    2004

  78. [78]

    pseudo-absence problem

    Poisson point process models solve the" pseudo-absence problem" for presence-only data in ecology , author=. The Annals of Applied Statistics , pages=. 2010 , publisher=

    2010

  79. [79]

    Biometrics , volume=

    Equivalence of MAXENT and Poisson point process models for species distribution modeling in ecology , author=. Biometrics , volume=. 2013 , publisher=

    2013

  80. [80]

    Diversity and distributions , volume=

    Predicting species distributions from museum and herbarium records using multiresponse models fitted with multivariate adaptive regression splines , author=. Diversity and distributions , volume=. 2007 , publisher=

    2007

Showing first 80 references.