This note describes the new features added to 'psimpoll' since my note discussing the original version (Bennett, 1992). The aims of the program are to:
1. 'psimpoll' is now written in C, conforming with the ANSI standard in every respect (as far as I can tell). This has certain disadvantages, notably with regard to screen handling, and one major advantage: it can be ported to any computer that has an ANSI C compiler. I have compiled and run the program as is under DOS, UNIX and Macintosh systems. This enables me to run it with the same dataset on different systems depending on the task in hand. For me, normal use is on DOS, but I use UNIX for analyses that are expensive in CPU time (such as time-series).
2. Age-depth modelling has been added, using linear interpolation, weighted curve fitting (by least-squares and singular value decomposition), spline fitting, and Bezier curve fitting. Calculated ages can be used to obtain accumulation rates from concentration data.
3. Confidence intervals are now available for pollen percentages (following Maher, 1972), and pollen concentrations (following Maher, 1981). Additionally, confidence intervals are assessed for age and accumulation rate estimates by simulation. The latter are combined with pollen concentration confidence intervals by error propagation techniques to obtain confidence intervals for pollen accumulation rates. The simulation process involves treating each radiocarbon age that actually was obtained as one estimate of the age from a population with the same mean and standard deviation as the measured one, and then repeatedly drawing "ages" from this distribution. The principle is that any of these randomly drawn "ages" is just as likely to have occurred as the ages that really were obtained. The random "ages" are then used to refit the age-depth model, and obtain confidence intervals for age estimates and sediment accumulation rates. The method works for any age-depth model, and produces results that are similar to exact calculation of confidence intervals in the relatively simple situations where this is possible. The procedure is discussed in more detail by Bennett (1994).
4. Zonation is now included, using least-squares and information content criteria for binary and optimal splitting, and also a version of CONISS (Grimm, 1986). A broken-stick model is used to help identify the maximum number of zones that should be recognised.
5. Other available analyses include rates of change (using any of eight dissimilarity coefficients), rarefaction analysis (Birks and Line, 1992), Fourier analysis by way of the Lomb periodogram for unevenly sampled data (see Press et al., 1992), principal components analysis (which also uses a broken-stick model to identify the usable principal axes: see Jackson, 1993), some basic statistical description of the data, and independent splitting, using the method of Walker and Wilson (1978).
6. Text on the diagrams can include any character from European languages based on Roman script, plus the Greek alphabet.
7. The 'psimpoll' PostScript output file consists of a series of pieces, which I term "boxes", of PostScript code. These "boxes" can be unpicked from the output file and combined with "boxes" from other output files to produce a new output file. This enables the selection and combination of parts of diagrams (scales and zonation schemes, as well as plotted curves), and is achieved through an ancillary program called 'pscomb' (pronunciation should rhyme with "tome", but "scum" will do).
8. There is a detailed manual (about 70 pages) that describes the working of the program and the principles underlying the various approaches and techniques.
My comments in the earlier note (Bennett, 1992) on my rationale for writing a program myself rather than using someone else's remain unchanged. It should be noted that the screen-handling is crude (the price of portability). The DOS- and the Macintosh-executable versions of 'psimpoll' are available, free and gratis, from Lou's pollen boutique by ftp. For other computers, you would need the source files (which I can send) and your own ANSI C compiler.
References.
Bennett, K.D. 1994. Confidence intervals for age estimates and deposition times in late Quaternary sediment sequences. The Holocene, in press.
Birks, H.J.B. and Line, J.M. 1992. The use of rarefaction analysis for estimating palynological richness from Quaternary pollen-analytical data. The Holocene 2, 1-10.
Grimm, E.C. 1986. CONISS: a FORTRAN 77 program for stratigraphically constrained cluster analysis by the method of incremental sum of squares. Computers & Geosciences 13, 13-35.
Jackson, D.A. 1993. Stopping rules in principal components analysis: a comparison of heuristical and statistical approaches. Ecology 74, 2204-2214.
Maher, L.J., Jr 1972. Nomograms for computing 0.95 confidence limits of pollen data. Review of Palaeobotany and Palynology 13, 85-93.
Maher, L.J., Jr 1981. Statistics for microfossil concentration measurements employing samples spiked with marker grains. Review of Palaeobotany and Palynology 32, 153-191.
Press, W.H., Teukolsky, S.A., Vetterling, W.T. and Flannery, B.P. 1992. Numerical recipes in C: the art of scientific computing. 2nd ed. CUP, Cambridge.
Walker, D. and Wilson, S.R. 1978. A statistical alternative to the zoning of pollen diagrams. Journal of Biogeography 5, 1-21.