Department for Environment, Food & Rural Affairs

Farm Scale Evaluations of GM Crops
2nd Interim Report

2 TESTING THE NULL HYPOTHESIS

2.7 Indicators of biodiversity

The development of biodiversity indicators during the early stages of the project has already been reported (November 1999 report, section 2.7). Since then, there has been further work on the protocols, largely to take into account the comments of the SSC, and also to take into account data from the 1999 field survey to re-assess sampling intensity at a given time. Field work on the 1999-2000 winter oil seed rape sites is also being used to pilot survey timings for the full scale programme starting in late summer 2000, and will be used to pilot the remaining untested field protocols.

The SSC has agreed with the contractors that earthworm sampling should not be undertaken as the agreed experimental design is inappropriate. Also, pilot studies on birds and small mammals will take place in 2000, but not within this project, and so they are not covered here. Nor does this report cover the monitoring of gene flow from the GM crops that is planned to take place.

For biodiversity monitoring, the order of which treatment is monitored first is decided at random. If this involves walking from a flowering GM crop to a non-GM treatment, then surveyors will change their overalls and boot covers to prevent pollen flow.

Samples are not being bulked prior to data entry, except where there are subsamples within a single sample point (e.g. soil seedbank sampling).

Here we concentrate on changes since our previous report. The major changes are:

2.7.1 Vegetation sampling in the crop

For seedling counts, median estimated coefficients of variation for total dicots and monocots per unit (half field) from the 1999 survey confirm that they fell within the range 20-25%. This would not be affected to an unacceptable degree by the reduction in quadrat size. For biomass, which was sampled using 1 m˛ quadrats, the coefficients of variation were higher, mostly 30-35%.

Biomass sampling has been modified. We found that frozen samples were, when de-frosted, unpleasant to handle and difficult to sort. This wasted time and lowered the quality of the results. Samples will in future be sorted fresh. The SSC recommended that an estimate of weed seed production should be obtained from biomass samples. We considered this recommendation carefully, but for the reasons stated in Section 2.7.3, opted to measure weed seed production more directly.

2.7.2 Vegetation sampling on the field margin

Vegetation recording on the field margin has been re-designed to coincide with three occasions when arthropods are sampled. On each of these occasions, a record will be made of (1) flowering species in 9 positions, (2) vegetation cover in 3 positions and (3) species seeding in 9 positions. Three types of marginal position are envisaged: boundary (always present), verge (unploughed, between plough edge and boundary) and margin (ploughed, either lacking crop or sown with grass). Verge and margin plots will be recorded only if the feature is >1 m wide.

The SSC recommended that we should attempt to record seed availability in the margin. A protocol has been written but was not available early enough in the season for it to be field-tested.

The following are therefore the main innovations:

2.7.3 Estimating seed return by seed traps

The SSC recommended that an attempt should be made to record weed seed return. It was suggested that this could be done from the biomass samples or on a per individual basis. However given the long period over which seed production occurs in arable situations and problems associated with estimation of seed production from biomass samples it was thought more accurate and representative to sample seed return regularly throughout the growing season. The protocol specifies ‘pot type’ seed traps which measure seed production on a per area basis which can then be related to the biomass samples.

2.7.4 Soil seed bank

To give an indication of potential differences among trial sites, preliminary analysis was carried out on soil samples taken from 7 of the fields used for protocol development in summer 1999. Based on this early experience, samples were taken from four winter rape sites in early September 1999. Three replicates of 1.5 l each were taken from 12 loci in each treatment, as follows:

4 farms x
2 treatments per farm x
12 sample points per treatment x
3 replicates per sample point = 288 samples (= 288 bags of soil).

Based on the initial flush of seedlings, seedbank abundance differed by a factor of about 13 from the largest to smallest. These data indicate that the minimal sampling intensity for estimating seed density and overall community structure is 6-8 samples from each treatment. This sample size is sufficient to detect a two fold difference between treatments in terms of seed density at the field centres. Examination of similar data from the literature and re-analysis of existing data, indicates that it is reasonable to expect changes in farm management practices to result in a difference of this order of magnitude.

2.7.5 Vegetation sampling in following crops

An attempt at follow-up sampling in autumn 1999 revealed that this is not the correct season for this operation. Fields may have been ploughed and show nothing. It is proposed that follow-up sampling should be done in June, when the follow-up weed flora will normally be fully developed.

2.7.6 Arthropods on vegetation

The objectives of this protocol are as follows:

Suction sampling will be undertaken by standard Vortis suction samplers. Although efficiency is always less than 100% suction samples represent estimates of population levels per unit area and thus allow statistical comparisons between treatments for the same habitat.

Sampling will be performed three/four times per year depending on crop growth phenology. Early samples (May), although potentially important in relation to crop management, are likely to contain a high proportion of immature plant bugs, which are not always possible to identify to species level.

The arthropods will be separated from other organic matter and soil particles by flotation, and then into different groups in the laboratory prior to being counted and identified as far as possible. Once separated, Heteroptera, Collembola and moth/sawfly larvae samples will be identified to species as far as possible, other groups to family or higher groups. A sub-sample of specimens (circa 6-10 per species/family etc) will be oven dried and weighed on a microbalance to provide an estimation of dry weight for that species/genus/family. Total biomass will then be calculated by multiplying up. A table of typical dry weights for each species/group/family will be compiled as the season progresses, and new species are encountered.

2.7.7 Sampling of crop pests

Although pests that occur on the crop do not contribute much to the biodiversity within the crop canopy directly, they may attract the attentions of parasites and predators, and there may be a potential interaction between those and the presence of other plants such as weeds, that are affected by the weed management programmes applied to the GM and non-GM halves of the fields. Therefore, an estimate of pest, parasite and predator abundance will be made at the same time that the Vortis samples are taken, i.e. once a month.

2.7.8 Further work required on field sampling protocols

The field protocols are virtually finalised, subject to fine-tuning during field training, and subject to final testing of the untested protocols on the 1999-2000 winter rape sites. The gastropod protocol needs to be finalised.

2.8 A review of the field studies, October 1999 – February 2000

The winter oil seed rape sites are proving valuable in protocol development, and will continue to be used in this way until harvest. A full report will be given later in the year. In summary:

Lincolnshire

Spital-in-the-Street
This was a paired field site. The weed flora, which was assessed on 3rd November, revealed that establishment of the GM crop had been patchy. Vegetation is due to be re-assessed in March. The conventional crop was treated with a graminicide on 5th October, while Liberty herbicide was applied to the GM crop on 3rd November. Arthropods and Carabidae were sampled on 15th October and are due, with the gastropods, to be re-sampled in April.

Croxby
This was a split-field site. Both the GM and conventional crops were treated with herbicide in November. Vegetation was assessed on 26th October is due to be re-sampled in March. Arthropods and Carabidae were sampled on 15th October and are due, with the gastropods, to be re-sampled in April.

Hertfordshire

Hemel Hempstead
This was a paired field site. The conventional crop was treated with a graminicide on 26th November and Liberty herbicide was applied to the GM crop in early February. Vegetation assessments were made on 1st November and are due to be re-done in March. There has been very poor establishment in some parts of the fields. Arthropods, Carabidae and gastropods were sampled on 15th October and are due be re-sampled in April.

[ Previous ] [ Contents ] [ Next ]

Published 13 June 2000
Farm-Scale Evaluations Index
Environmental Protection Index
Defra Home Page