Recommendations for depletion modelling of granivorous birds
|
|  Yüklə 6.33 Mb.
|
IntroductionRelatively few studies on farmland bird ecology aim to measure food availability. As discussed under Objective 1, availability is a function of the abundance and accessibility of this food, as well as the foraging behaviour of the bird species (food and habitat preferences and response to changes in food availability). In practice, many studies measure relative (rather than absolute) food abundance. Although this undoubtedly simplifies matters (the methods used need not be efficient), the sampling method(s) utilised can influence the reliability of the data obtained. The method used is generally governed by the diet of the species of bird being investigated. Most sampling methods are very selective, and have a wide range of effectiveness (Hutto 1990; Tones et al. 2000). Sampling can thus over-represent, under-represent, or not represent at all, the food abundance in an area. In addition, each method has its associated assumptions and limitations, and each will differ in terms of repeatability and hence reliability of the estimates obtained. If data from different studies are to be compared, or indeed combined in order to build the models being considered in this project, some degree of standardisation (of both methods and measures of abundance) seems necessary. Bird species may be separable into guilds sharing common diet elements, and so may be investigated using similar techniques, although one single method is unlikely to be suitable for all circumstances. For example, methods of measuring arthropod abundance are likely to be more or less efficient for different species depending on mobility and size. Different methods, also, are likely to differ in terms of repeatability.
To: (i) review the literature on sampling methods used to estimate density or biomass of bird food/prey items; (ii) assess repeatability of the estimates obtained using the different methods; (iii) assess to what extent the methods measure food availability to the relevant species; (iv) assess to what extent it is necessary to standardise the methodology for assessment of bird food resources, in order to include such data into the proposed models; and (v) make recommendations on the methods that should be used.
6.3.1 ArthropodsThe standard techniques for measuring the absolute and relative abundance of arthropods are listed in Southwood (1978), which remains the definitive text on the subject. Powell, Walton & Jervis (1996) list some of the most commonly used techniques and describe their limitations. The techniques relevant to the estimation of the abundance of insects are shown in Table 6.1. Calibration, where indicated, is necessary for the estimation of absolute abundance. Table 6.1 Field-sampling methods for estimating insect abundance (derived from Powell, Walton & Jervis, 1996).
Sunderland et al. (1995) review the principle methods available for the estimation of invertebrate predator densities (which include many listed in Table 6.1) and discuss the relative advantages, disadvantages and limitations of each. They conclude that there is no single method that is suitable for all circumstances, and recommend methods according to the types of predators being assessed, the habitat and the scale of the investigation. Similarly, the methods used to estimate the abundance of arthropod bird foods should be appropriate to the species of arthropod being measured, and to the foraging habitat and behaviour of the birds being considered. Most of the studies described in this chapter select techniques primarily on this basis, rather than with regard to standardisation of methodology. This has resulted in a wide range of techniques being used to estimate the abundance of invertebrate food for farmland birds and may reduce the potential for between-study comparisons. Stewart & Wright (1995) list a variety of vacuum nets (suction samplers) which have been designed for extracting insects and other arthropods from natural vegetation and agricultural crops. Macleod et al. (1994) describe a modified petrol-driven leaf-collecting device, and compare its efficiency to that of the ‘Thornhill vacuum sampler’ (Thornhill, 1978), and Arnold (1994) describes a lighter and easier to use device, the ‘Vortis’ suction sampler. The Game Conservancy Trust (GCT) have generally used a Dietrick vacuum net (D-Vac) suction trap, which is similar to the Thornhill device (Dietrick, 1961), although Stewart and Wright (1995) show that their ‘Blow & Vac’ device achieves a greater air velocity and hence improved efficiency of extraction, although it has a smaller sampling head. A study of D-Vac efficiency found variable capture rates of between 43% and 70% for ground-dwelling linyphiid spiders, compared to capture rates of 84% to 92% for linyphiids that lived higher in the crop (Sunderland & Topping 1995). Vegetation density and species microhabitat preferences influenced capture rates. Mommertz et al. (1996) concluded that short-term D-Vac sampling is not appropriate for Carabidae, Staphylinidae and Lycosidae, because relatively large individuals are under-estimated. In grassland, Standen (2000) found that both pitfall traps and D-Vac/sweep netting were required to adequately sample species richness. This was because many species were “method-unique”. The published literature on farmland bird ecology was then examined, and studies in which bird food availability or abundance was measured were reviewed. Using a D-Vac, Ewald. & Aebischer (1999) took five ten-second sub-samples, each of 0.092 m2 along a diagonal transect into a field. Most invertebrates were identified to the family level, some to genus or species. Holland et al. (2002) used a similar method to estimate invertebrate chick food for farmland birds, as did Moreby et al. (1994) to compare the abundance of invertebrates between conventional and organic winter wheat fields. The D-Vac method was also used by Moreby & Southway (2002) to measure the availability of invertebrate groups important in the diet of nestling farmland birds, in order to investigate crop preferences in arable farmland, and by Moreby & Southway (1999) to investigate the influence of herbicides on autumn food available to birds. In a study to assess the indirect effects of pesticides on birds, Morris (unpublished draft report to DEFRA) used a petrol-motor leaf suction device (Stihl BG 75) equipped with a 10 cm diameter suction-hose, rather than a D-Vac to sample invertebrate food for yellowhammers Emberiza citrinella , because the latter required two operators. He found that catches depended on weather conditions and the height and density of vegetation. Therefore invertebrate populations from patches with different vegetation structure could not be compared. However, he argues that the results provide a reasonable indication of relative prey accessibility (which is a component of prey availability – see Objective 1). He also used a sweep net (a kite net – as suggested by Evans, 2001) to sample aerial invertebrate food for swallows Hirundo rustica, and a 10 x 10 cm soil corer to sample abundance and biomass of earthworm food for lapwings Vanellus vanellus. In a study of the effects of spraying strips along crop edges on non-target insects, de Snoo & de Leeuw (1996) estimated insect abundance by sweep netting. In each 100m strip, 10 sub-samples were taken in six sweeps of a sweep net with a diameter of 35 cm. The total area sampled was 20 m2 per 100m. Sampling took place 1.5 m from the field edge. For mobile insects, additional visual observations were made. In his study of grey partridge Perdix perdix and red-legged partridge Alectoris rufa chick ecology, Green (1984) took two sets of five 0.1 m2 D-Vac samples, 5 and 50 m from each field boundary. In cereal fields, fifty sweeps were made with a 1.0 x 0.4 m sweep net, 5 m from the field boundary. Each sweep covered c.1 m and a pace was taken between each. All sweep netting was done by the same person. In an experiment to test whether pesticides reduced the survival of grey partridge chicks, Rands (1985) measured insect food availability using a sweep net, taking 50 sweeps per field edge. Hill (1985) measured insect food supply of pheasant Phasianus colchicus chicks by taking five D-Vac samples and five 50 m sweep net samples within the brood range during their second week. Between-year variation in insect abundance was estimated from D-Vac sampling data. Beintema et al., (1991) measured prey abundance for breeding wading birds by sweep netting and pitfall trapping. Ten pitfalls were used, in two rows of five, at 10 m intervals. At 12 points between the pitfalls, a sample was taken with the sweep net, each time making three sideways sweeps when walking. A suction apparatus, ‘based on a small fan, sucking air through a narrow nozzle’ was also used to remove as many insects as possible from a 50 x 50 cm quadrant. In a study if the breeding ecology of farmland yellowhammers, Stoate et al. (1998) used both the D-Vac method and sweep netting to sample ground-dwelling invertebrates. Five 0.5 m-2 D-vac samples and five sweep samples, each of ten sweeps, were taken at each sampling location. Sweep netting was also used in a study by Brickle et al. (2000) to sample the abundance of corn bunting Miliaria calandra chick foods around nests in the GCT Sussex study site in 1996-97. Twenty sweeps were made during the first week in July in each habitat block. For large fields the mean of up to three samples was used. Barker et al. (1999) used sweep netting and emergence traps to estimate sawfly abundance in grassland and arable fields, and in cultivated vs. uncultivated experimental plots. Møller (2001) estimated swallow food abundance on farms with and without dairy cattle using sweep netting. Nine transects of 20 m in each 40 ° direction from the farms were sampled, on pasture. |