I am reading the book "Occupancy estimation and modelling". The occupancy model and ecological niche modelling are similar in the ability of estimating species distribution. However, I feel there is some difference between them, which I could not clearly figure out. Please share your ideas. Thanks.
One difference is that with the occupancy models you can estimate the prevalence of the species
Occupancy may defined as the occupied niche in contrast to the fundamental niche.
Here you can find some courses about occupancy. I'm sure you can use them.
http://www.mbr-pwrc.usgs.gov/workshops/
I think of occupancy models as models in which you use observed presences to infer habitat suitability over a grid of cells, usually where you've measured likely predictor variables (e.g., rainfall, temperature) from other GIS layers. By contrast, niche models should ideally be based on a more mechanistic understanding of the organism's physiology, habitat needs, interspecific interactions, etc. - i.e., you don't need to know where the organism occurs in order to produce a niche model, just what it needs (and tolerates). If both models were perfect they would produce the same predictions, but in practice, I have issues with assuming that they are identical. People often use the two terms interchangeably so the peer-reviewed literature can be quite confusing.
The occupancy models were developed to solve the problems created by imperfect detectability. These models use information from repeated observations at each site to estimate detectability. With them you can estimate at the same time the occupancy and the detectability. If the detectability can be near "1", the results from a occupancy models and ecological niche modelling would be the same.
So occupancy models require repeated observations over time for each site? What about a situation where we have data on presence/absence of parasites within a host species and would like to predict where the parasite can occur given environmental correlates? Any suggestions on which approach to use?
If you don't have a problem with the detectability (is always =1), you could try to use a GAM, GLM or a species distribution model program like MaxLike. You have this package in R:
http://cran.at.r-project.org/web/packages/maxlike/maxlike.pdf
One important difference is that occupancy models account for false absences during the sampling, whereas the niche modeling deals only with presence data, assuming that the species do not occur where it was not registered. An interesting thing to point out is that the occupancy estimates (i.e., the site estimates) can be used to model the niche distribution of a given species while also accounting for imperfect detection and false absences.
actually occupancy model will used for data,that you are not sure to identify sex or ageclass or demographics.
Hello, I think one could consider occupancy models as another algorithm (correlative models) that can be used for ecological niche modeling, just as GLMMs are. The big difference, as the others point out, is the type of information that the occupancy models require, which consists of standardized sampling with repeated visits. The problem with these types of models is that they pose a logistical difficulty when working at large scales, as is usually done in niche modeling (or distribution), because it is necessary to capture the heterogeneity in space M (dispersion capacity of the species), that is, we would need standardized sampling throughout large part of the area to obtain a robust prediction. If we have sampled a good part of the area and the probability of detection in the occupancy model is almost 1, we expect similar results to other algorithms (if the detection is lower, we probably have even better results than other algorithms).
Another relatively new approach that follows the logic of hierarchical models (such as occupancy models) is that of integrated species distribution models, which use data from both standardized sampling (detection / non-detection), and presence only data, that are generally used in classical niche algorithms. You can find more information at: Dorazio et al (2014)
Article Accounting for imperfect detection and survey bias in statis...
Koshkina et al 2017 Article Integrated species distribution models: combining presence-b...
Schank et al 2019
Article A Sensitivity Analysis of the Application of Integrated Spec...
Fletcher et al 2019
Article A practical guide for combining data to model species distributions
- Badges
- Science topic
- Similar topics
- Engineering
- Modeling