Worcestershire Record No. 16 April 2004 pp. 42-45

Terry Knight

The reserve
Descriptions of the reserve and its management are given in Knight (2003a, 2003b). A plan and typical cross section are shown  in Figure 1.

Previous recording
A few fixed points were set up from the early 1980's to monitor changes in the grassland vegetation. These consisted of 1 metre square sample areas each divided into twenty five 200mm by 200mm compartments. Because of the time taken to fully record these points only five have been so far been established.

Alternative methods
For sampling grassland vegetation in order to classify the type of plant community a few 2 metre square sample areas were used with the points chosen where the vegetation appeared to be homogeneous (Rodwell 1992). Whilst this method and the 1 metre square method above can give a good assessment of the plant communities in a few selected areas, because of the variable nature of the reserve grassland, they are unlikely to be able to characterise the whole of the grassland. In order to monitor changes in the composition of the grassland it was considered that a large number of sample areas of smaller size scattered across the reserve might be more useful. It seemed sensible to make these 200mm by 200mm to match the compartment size used in the 1 metre square samples. However, the England Field Unit has used 100mm by 100mm sample sizes for some of its monitoring. A more complicated method has been used by Sheffield University in which each sample area was 1 metre square but divided into only six compartments of varying size nested in one corner. These two methods are described in Crofts and Jefferson (1999).

The small size sample areas have advantages over the other methods. Because many more areas can be sampled statistical methods can be applied to the results and hence future recording does not depend on finding the same points. It is only practical for one person to record two 1 metre square areas in a day and all twenty five compartments have to be done at a time. In the scattered 200mm by 200mm samples it only takes less than ten minutes to do each one and the recording process can be stopped and started at any time. Recording the twenty five compartments of the 1 metre square is found to be much more tedious and physically tiring, particularly reaching the centre compartments, than in the many small samples.

As the reserve grassland is fairly rank and dominated by Tor-grass, a sample size less than 200mm is probably unsuitable as few taxa and plants would be recorded in any sample whereas in a fine downland turf many more would be.

Method adopted
It was decided to use the 200mm size sample area and so a stout wire coat-hanger was bent to form a throwing frame with a handle as shown in Figure 2. To reduce recorder bias the frame was thrown over the shoulder in a southwards direction at each sample location to give an actual sampling point. The only exception to this was when, by so doing, it was likely to land in a patch of thick scrub. When this was likely, fairly rarely, the angle of throw was altered to miss the scrub.

To ensure there were sufficient samples for statistical analysis a minimum of 36 was used in each of the four paddocks and each of the four north-south lines which amounted to at least 144 overall. The actual totals were 151 in 1998 and 147 in 2002. The paddocks sampled were Paddock 1 at the north end and Paddocks 3, 4 and 5 at the south end (see Plan). The four lines were Line E along the top of the paddocks, Line EC just above the central path, Line WC just below the central path and Line W along the bottom of the paddocks (see Typical Cross section). This gave a more or less even grid layout over the whole of the grazed area.

Sampling was carried out between late July and mid August when most herbs had above-ground parts and seedlings were advanced enough to attempt identification. However, difficulty was found in particular in distinguishing between the following combinations of small seedlings:

No attempt was made to distinguish between species of the critical groups, Dandelions, Roses and Brambles.

At each sample point the number of plants of each species having their main root within the frame was noted. For creeping plants with many stems, estimating the number of plants was difficult and the best that was achieved was a subjective assessment of the number of stem clusters. It is likely that the number of plants was over-estimated for species such as Field bindweed.

Grasses and sedges
In 2002 seven relatively easy to identify species were selected and their presence or absence in each sample was noted.

Randomness test
Although the exact point of a sample was determined in a random fashion at each location, the locations were on a non-random ,or systematic, grid layout. In 2002 a test was undertaken in Paddock 1 when it was sampled again using locations pre-determined from random number tables. When the results were compared it was found that there was no significant difference in the mean of the number of species or the number of plants obtained from the test method and the selected method.

RESULTS
The total area sampled in 1998 was 6.04 sq. metres (151 samples) and in 2002 was 5.88 sq. metres (147 samples) which aggregates to 11.92 sq. metres (298 samples). 68 taxa (representing at least 68 species) and 2725 plants which were not grasses or sedges were recorded from the aggregate area. A full list of these 68 taxa with their frequency and abundance in the 298 samples is given in Table 1. The twelve most frequent and twelve most abundant are listed in order in Table 2. The frequency in 147 samples of the selected seven grasses and sedges given in Table 3.

In Table 1 the estimated total number of plants is derived from a simple proportioning by area using the formula (total number found x 25 x 31950 / 298) where the area of grassland in the four paddocks is taken as 3.195 ha. This is likely to be very inaccurate for the least abundant species. The average number per square metre in the table is calculated from the total number found divided by 11.92.

Eleven 1 square metre areas were derived by aggregating blocks of 25 samples. This gave the number of herb species per square metre as at least a minimum of 18, an average of 27.5 and a maximum of 32. Adding the most likely numbers of grasses and sedges gave totals of a minimum of 26, and average of 37.5 and a maximum of 42 per square metre

The five species (Common knapweed, Common bird's-foot-trefoil, Burnet saxifrage, Lady's bedstraw, and Meadow vetchling) which were found to be clearly the most frequent also were found to be the most abundant. For these five the frequency varied from 32.6% to 26.3% of samples and the average number of plants per square metre was 22, 35, 16, 27 and 15 respectively. However, the average number of plants per sample where they were present works out as 2.7, 4.3, 2.1, 3.7, and 2.2 respectively. This suggests that Common bird's-foot-trefoil and Lady's bedstraw are more likely to occur in patches whereas the other three are more likely to occur as isolated plants.

Further analysis
The above results are obtained without using statistical methods. From work done so far it seems that statistical analysis can provide much more information, differences up the slope and along the slope for example. The data collected should also establish a baseline from which future trends can be determined.

CONCLUSIONS
Results from the selected method indicate that it is a simple way of collecting statistically useful data in grassland that is not too rank. It is not dependent on re-locating former sampling points and requires no equipment apart from the small frame. Although it took one person about twenty hours to do the survey in each of the two years this is not considered to be too onerous.