DENSITY 5.0 Frequently asked questions

1. How many captures and recaptures do I need to use DENSITY?
   It is most critical to get enough recaptures. To estimate density by SECR you will probably need at least 10 recaptures, and 20 is a safer minimum. AIM FOR MANY MORE THAN THE MINIMUM! Precision improves rapidly with additional recaptures, with slowly diminishing returns over 50 or so (e.g. Efford, Dawson & Robbins 2004 Fig. 2).  Some studies have many recaptures of few animals (<20 individuals); estimation is possible, but confidence intervals will be very wide. To fit more complicated models with covariates, finite mixtures, trends etc. you will need more data.
2. I cannot open the help file. What is wrong?
   First check that there is a file Density5.chm in the folder with Density5.exe. Density5.chm is a compiled Windows help file, created with the Microsoft HTML help compiler version 4.74. It is seen by the operating system as an executable file, and treated by some systems as a security risk. This is the most likely cause of your problem. Try opening the help file from Windows Explorer and overriding any security warning. If you downloaded a 'naked' version of the help file then right-click on it, select Properties, click Unblock and Apply.
3. Should I use R?
   Almost certainly, yes. The R package secr greatly extends DENSITY. Many more models can be fitted, and fits may easily be compared by AIC. Spatial variation in density may be related to habitat variables. Model averaging is enabled. Complex sequences of data manipulation and analysis may be prepared as R scripts. R can be impenetrable on first encounter, so we provide the R interface in DENSITY as a bridge. The interface can  translate most of what you have been doing in DENSITY into R code. This excludes simulation, for which the DENSITY Simulator is probably more powerful.
4. What about Bayesian methods?
   Andy Royle and collaborators have recently shown how data augmentation and Markov chain Monte Carlo methods may be used to fit some SECR models (e.g., Royle & Young (2008) Ecology 89: 2281-2289; see also Marques et al. (2011) and Efford (2011)). So far, these are largely a subset of what is available in DENSITY and 'secr', but the approach has potential advantages for modelling random effects.  Sceptics will note the challenge of model selection. Bayesian methods perform no better than the methods in DENSITY and can be much slower.
5. Why are the confidence intervals for estimated density so wide?
   Perhaps you don't have enough data.  More constructively: What is the basis for your comparison?  If you are tempted to use N-hat/A-hat, make sure you include uncertainty in A-hat as well as N-hat (see Calculator on 'ETA density tab').  Also, the default intervals reported from IP SECR and ML SECR are for the expected population density; that is, they include spatial process variation in population density that is 'conditioned out' of conventional ETA estimates (N-hat/A) (but not estimates by conventional distance sampling). There are three ways to get more comparable intervals from SECR (i) in Options | ML SECR set Distribution to Binomial, (ii) use an estimate of the precision of realised population size from Efford and Fewster (2012; see also 'secr' function region.N) or (iii) from IP SECR use the Adjusted SE (see Help index).  And remember that your estimate of N-hat will have spurious precision if it does not model individual heterogeneity due to home range location - this is automatic in SECR. 
6. The font in the Help file looks heavy and blurry. What can I do?
   Open Internet Explorer and under Tools | Internet Options | Advanced | Multimedia switch off 'Always use Clear Type for HTML'. (Windows uses IE when displaying .chm files)
7. Can I pool data from different grids?
   Yes. Set up the capture file with the data for each grid as a separate 'session'.  In ML SECR pooling is achieved by defining a 'between-sessions' model in which a parameter is constant across sessions.  For example, a single detection parameter may be fitted across multiple grids.  There is no facility for estimating empirical (between-grids) encounter rate variance. For IP SECR, the same trap layout must be used for each session. Use a Session filter such as [1-3] to produce a combined estimate. Usually you will want to use Options | Input 'Separate ID each pooled session' and Options | Output 'Re-scale density from pooled sessions' (both of these are on by default). 
8. Can I manipulate the design matrix, as in MARK?
   No. DENSITY packages the ML SECR models you are likely to need so that you don't have to wrestle with design matrices. You define models in DENSITY mostly by clicking options and entering covariates. The open-ended style of MARK (and RMark) is very attractive, but we felt that novice SECR users needed a tighter and more prescriptive approach. A much more extensive model set is available in the R package secr; the R interface provided in DENSITY 5.0 should get you started.
9. How can I run CAPTURE?
   First you need the program - you can get it from the Patuxent software archive maintained by Jim Hines. In 'Options | Closed N' browse to the location of the CAPTURE executabl and download it to the folder with Density5.exe. Now you can run CAPTURE from the CAPTURE tabbed page on the right of the main screen (assuming you have loaded some data). Right-click on the task window to select tasks from a popup menu.  Be warned that CAPTURE can choke on very long file names, and there is a limit on the number of animals (at least in the Patuxent version).
10. How should I be guided by the model selection algorithm in CAPTURE?
    For non-spatial analyses you should be aware that the algorithm lacks power and may mislead when data are sparse. For spatial analyses (IP SECR or ML SECR) there is the added complication that the tests in CAPTURE assess individual heterogeneity caused by animal location as a violation of the null model, while the spatial 'null' model expects this heterogeneity and allows for it. Nevertheless, test results from CAPTURE showing a strong learned trap response should be taken seriously.
11. Why does the program sometimes lock up?
    I don't know. The interface has a few low-priority bugs that surface mostly when exotic (over-ambitious) models are fitted to small datasets. Please report and move on :-). Ctrl-Alt-Del will bring up the Windows task manager so you can end the Density process.  Count to ten and remember what you paid for it...
12.  Can I display  trap sites overlaid on a GIS layer (raster map)?
    Yes. See 'Background image' in the help index.
13. What scale limitations are there in DENSITY?
    In principle, none. In practice, the interface works best for trap spacings in whole metres.
14.  How should I cite DENSITY?
    The recommended citation for the present version of the software is
    Efford MG 2012. Density 5.0: software for spatially explicit capture-recapture.  Department of Mathematics and Statistics, University of Otago, Dunedin, New Zealand. http://www.otago.ac.nz/density.
    Citations for specific SECR methods are in the References (see the Publications page on the website for the latest).