There are a few tricks to estimating cover. I suggest you start by looking at the California Native Plant Society guidelines for vegetation assessment. They have a cover diagram which helps to estimate different categories of cover.

http://www.cnps.org/cnps/vegetation/pdf/percent_cover_diag-cnps.pdf

The first rule is: the cover of any one thing, and the total cover, is usually less than you think.

The main technique I use for estimating cover in a 1m2 quadrat is to remember that every 10x10cm square is 1%. So, looking at each species, you can rapidly count up the individual covers of each plant (where a single erect stem with little horizontal leaf cover is, say, 0.1% cover). After a while you'll be able to estimate cover by eye without adding up individual plants, using either Braun-Blanquet cover classes, or actual percentage cover (in small quadrats, I prefer to estimate actual percentage cover).

So, for your liana, the question becomes, how many 10x10cm squares does it cover? Use half and quarter squares for parts where it does not cover a full 10x10cm square.

Hi Evans,

1x1m is relatively small, so it would be rather easy to take a photograph of the quadrats,and then simply overlay it with a grid and get a very good estimate. An example is http://pubs.aina.ucalgary.ca/arctic/Arctic63-3-315.pdf

If larger plants overlay smaller ones this still works - only you will of course need to estimate the cover of the lowest vegetation layers directly in the field.

You might also want to look at the methodology of GLORIA: http://www.gloria.ac.at/ 

Thank you very much you all for your advice!

As Alan Short said you should divide your quadrat into 10 x 10 cm squares. You should ask yourself when you determine a cover percentage for a plant why it is not up or down what you think its right? after few measurement, you will see that you can determine the cover percentage more proficinally

(1) It's a tradeoff between quality and quantity of information.  The smaller the squares, the higher the quality - but the downside is that with smaller squares you will have time for fewer plots and/or fewer squares/plot. 

(2) Some bad news is that the size of the grid cells affects the estimate. I don't have the references for this handy, but some of them go back to the 60's and 70's. 

(3) The least biased method  is "optical point cover" in which you record the cover at a set of "infinitely small" squares (= optical points) - which in practice generally means the intersections of cross-hairs (such as crossed threads) - this what all the other methods are compared to.  I have used this method.  With a well-built threaded frame, it can be surprisingly fast.  And no need to worry if your visual estimates are drifting too high or too low.

(4) The smaller the square, and especially with optical points, the more important it is to have something to deal with parallax.  A standard solution is two grids, lined up, with one ~ 5-20 cm above the other. If you visually line up the grid points, and then sight down to the plants, you will have unambiguously defined a point or grid cell.

(5) One method that some US range scientists have used is to have "standardization plots" located at their field headquarters. Cover is measured very accurately, very frequently, in these plots.  Everyone "re-calibrates" their visual cover estimates every morning in these plots before going out to collect data.

(6) Statistically, almost any measure of cover is easier to analyze with standard statistical models than cover classes.  Even if you assign each class into its median value (e.g., 25% - 50% becomes 37.5%), there are too few possible values to assume a continuous variable (hence we can't use anova or linear regression on these values), not to mention the residuals are almost certainly not going to be normal.  If you treat cover class as a categorical variable you are losing a lot of information.  This is another argument for scoring lots of small cells, or optical points, as yes/no for each species and adding up the 'yesses'.

Hope this is helpful.

Norma Fowler

Dear all,

I would like to advance in this question. As Evans, i have a doubt. My research is in scrub vegetation where clonal growth is prevalent.

I want to determine plant cover in 1 m2 quadrats, but the literature is extensive (optical point cover, visual estimation and pin-point cover (Jonasson, 1988)). In your opinion, which method can be more accurate? I am tending to pin-point cover, but i am not sure.

One of the most important things to do before you start weeks of (generally expensive) field work is to do some preliminary studies and a literature review to optimise your sample size, sampling methodology and precision.

But before you start on the preliminary study, you should have a clear idea of your objectives and hypotheses, as well as HOW you intend to analyse and interpret your data.

The pilot study is overlooked far too often, with researchers picking methods and sample sizes that have been used in previous literature (often without pilot studies themselves) and simply repeating those methods. So, to use some of the examples given above, you could

1) Do increasing numbers of quadrats (up to 100), estimating cover in each quadrat by two or three methods and then comparing the cumulative precision, accuracy and repeatability of each method until you have a defendable reason to pick your method and your sub-sample size (sub-sample size in this case means number of quadrats per plot - however, the number of PLOTS is often more important. If you are doing multiple plots then the plots become the samples and the quadrats become the sub-samples; you need to beware of pseudo-replication).

2) You could also examine other methods like nested quadrats to simply look at a quadrat frequency - quadrat frequency is accurate, fast and repeatable but gives you a different class of data from cover. There are also methods like Whittacker plot,  or just point-canopy intercept (without the quadrats), line-intercept, and many others. If you are looking at the effect of some environmental factor on your vegetation, it may be more useful to do many rapid (but less precise) Braun-Blanquet plots to get a larger sample size (number of plots) than a smaller but more precise sample size.