11.Work Needed to Extend the Range of Bird Species That Can Be Covered by Deterministic Models

11.Work needed to extend the range of bird species that can be covered by deterministic models

11.1Introduction

The deterministic (empirically-based) modelling approach requires information on the relationships between: a) food abundance and breeding parameters and (b) breeding performance measures and population status, for a range of farmland bird species. It is known that this information exists for the grey partridge (Perdix perdix). Further work has recently been completed on a number of other farmland bird species. A review of this work is therefore needed in order to identify the data available, and assess it in terms of its usefulness for the model.

11.2Aim

To produce a matrix of declining farmland bird species and datasets describing food availability, annual productivity and population change, to assess the suitability of the datasets for the analytical framework proposed under Objective 10, and to indicate the studies required to fill the gaps.

11.3Methods

11.3.1Literature search

Published papers were derived from previous reviews on the indirect effects of pesticides on birds (McKay, unpublished) and the impact of herbicides on weeds (Marshall et al., 2001). In addition, Biosis was searched to find any papers published subsequent to these two reviews. Additional information from the ‘grey area’ was obtained from the consortium’s knowledge of recent and on-going studies. Bird species considered were the 21 identified by the consortium at the start of the project.

11.3.2Evaluation of data sets

Each study was examined to determine the type of data collected, the sampling method used and the parameters included in any population model.

11.4Results

11.4.1Datasets available

Table 11.1 lists the species being considered in this report, and for each, the studies which have attempted to measure food availability and breeding performance (intensive autecological studies) or model population breeding parameters and population size (demographic studies).

Table 11.1Farmland bird species and published papers relating breeding performance to either food availability or population size. The food availability, productivity and population parameters used in each study are listed.

Species / Reference / Food availability / Annual productivity / Population change
Chaffinch
(Fringilla coelebs) / Siriwardena et al. (2000) / clutch and brood sizes, chick:egg ratio and daily nest failure rates / population decline
Cirl bunting
(Emberiza cirlus) / None found
Collared dove (Streptopelia decaocto) / None found
Corn bunting / Brickle et al. (2000) / invertebrate abundance / nest survival and weight of chicks
(Miliaria calandra) / Aebischer & Ward (1997) / densities of food items / density of nesting birds
Siriwardena et al. (2000) / clutch and brood sizes, chick:egg ratio and daily nest failure rates / population decline
Goldfinch
(Carduelis carduelis) / Siriwardena et al. (2000) / clutch and brood sizes, chick:egg ratio and daily nest failure rates / population decline
Greenfinch
(Carduelis chloris) / Siriwardena et al. (2000) / clutch and brood sizes, chick:egg ratio and daily nest failure rates / population decline
Grey partridge / Green (1984) / densities of food items / chick survival
(Perdix perdix) / Potts & Aebischer (1991; 1994) / fluctuation in the food supply / chick mortality rate / annual variation in population density
House sparrow
(Passer domesticus) / Freeman & Crick (2002) / breeding performance / population changes since 1975
Linnet
(Carduelis cannabina) / Siriwardena et al. (2000) / clutch and brood sizes, chick:egg ratio and daily nest failure rates / population decline, 1975-1986
Moorcroft & Wilson (2000) / egg and nestling loss, clutch size, brood size and fledglings per attempt / population size 1962-1995
Quail (C. coturnix) / None found
Table 11.1 continued.
Species / Reference / Food availability / Annual productivity / Population change
Red-legged partridge
(Alectoris rufa) / Green (1984) / densities of food items / chick survival
Reed bunting
(Emberiza schoeniclus) / Siriwardena et al. (2000) / clutch and brood sizes, chick:egg ratio and daily nest failure rates / population decline
Rook
(Corvus frugeligus) / None found
Skylark
(Alauda arvensis) / Chamberlain & Crick (1999) / clutch size, brood size and post-hatching survival rate / population decline
Siriwardena et al. (2000) / clutch and brood sizes, chick:egg ratio and daily nest failure rates / population decline
Stock dove
(Columba oenas) / Siriwardena et al. (2000) / clutch and brood sizes, chick:egg ratio and daily nest failure rates / population decline
Stone curlew
(Burhinus oedicnemus) / None found
Tree sparrow
(Passer montanus) / None found
Siriwardena et al. (2000) / clutch and brood sizes, chick:egg ratio and daily nest failure rates / population decline
Turtle dove
(Streptopelia turtur) / Browne & Aebischer (2001) / number of young fledged per pair / annual population change
Siriwardena et al. (2000) / clutch and brood sizes, chick:egg ratio and daily nest failure rates / population decline
Woodpigeon
(Columba palumbus) / None found
Yellowhammer
(Emberiza citrinella) / Siriwardena et al. (2000) / clutch and brood sizes, chick:egg ratio and daily nest failure rates / population decline
Yellow wagtail
(Motacilla flava) / None found

11.4.2Evaluation of datasets

11.4.2.1Corn bunting

Brickle et al. (2000) sampled the abundance of four of the main corn bunting chick food invertebrates: Opiliones, Lepidoptera larvae, Symphyta larvae and Orthoptera. These invertebrates were found to account for the majority (62%) of chick diet (Brickle & Harper 2000). Invertebrates were sampled by sweep netting, but it was noted that this did not sample an important component of the diet – large lycosid spiders, and resource constraints prevented additional techniques being used. Invertebrate abundance was measured at foraging locations within the territory around each nest. The measures of breeding success used were clutch and brood sizes, chick weight and nest survival. The authors were able to regress food availability against residual mean chick weight (slope = 0.11 ± 0.005, t24 = 2.21, P = 0.036). In conclusion, this appears to be a very useful data set where effort has been made to ensure that the chick food sampled was that available to the foraging birds, and to identify possible biases and sources of error.

Aebischer & Ward (1997) examined the diet of nestling corn buntings in their study area by faecal analysis. Data on invertebrate abundance (three taxa known to be important in the diet: large beetles, caterpillars and spiders/harvestmen) was obtained from previous sampling within the study area, but was not directed towards individual territories or nests. Abundance measures were averaged over five years. Sampling was by D-Vac of spring- and autumn-sown cereals. The efficiency by which this method samples the three target taxa is not discussed, but is likely to be low for the larger and more mobile species. Breeding success was not measured directly, but singing male corn bunting density was measured. The authors were able to regress components of chick food (density of food items) against corn bunting density. In conclusion, this study is more broad-based than the previous one and as such is of less use, particularly as no direct measurement was made of breeding performance.

Siriwardena et al. (2000) modelled breeding performance (BTO nest records data) and population decline (BTO Common Birds Census data) and found no evidence of a relationship (but see comments in Discussion section).

11.4.2.2Grey partridge

Green (1984) investigated the diet of grey partridge chicks using faecal analysis and found it to be composed of various invertebrate taxa and grass spikelets. He then sampled the abundance of these foods during the main hatching periods, within the foraging areas of the birds (which were radio-tracked). D-vac sampling, sweep netting and ‘quadrat weed samples’, were used. Chick survival rate was measured. The authors were able to correlate chick survival rate with the abundance of commonly eaten arthropod prey. In conclusion this is a useful dataset as some attempt has been made to sample the invertebrate and plant food items available to foraging birds and a sample size of 13 farms was used.

Potts & Aebischer (1991), in an extension to work first presented in Potts (1986), used a wider variety and range of data on the abundance of grey partridge chick food in the GCT Sussex study site, in their model of the population dynamics of the grey partridge. Field-by field measures of the abundance of insect food from 1970 were used, as were results of intensive studies of the effects of pesticides on this food from 1973. Their insect abundance index was obtained by summing the densities of five insect taxa: small beetles, sawfly and other caterpillars, leaf beetles and weevils, plant bugs and leaf hoppers, and aphids. The sampling methods used are described in Green (1984). They recalculate relationships between chick survival and the index of insect abundance, for five farms in the Sussex study area, and for 17 radio-tracked broods. In conclusion, this is a useful extension to the work reported in (Green, 1984), but mainly uses the same dataset.

Potts & Aebischer (1994) construct a population model based on data from the GCT Sussex study site as well as from two sites in Norfolk, and demographic data from 36 populations from nine countries and the national UK data from the GCT’s National Game Census. They use this model to show that annual fluctuations in the population can be attributed to annual variations in chick survival rate.

11.4.2.3House sparrow

In a recently completed study investigating the decline of starlings and house sparrows in Great Britain, Freeman & Crick (2002) model the effects on the population of changes in breeding performance measured using the BTO Nest Record Scheme data, and changes in adult and first year survival, measured from ringing recoveries. They found that the best fit to the data was found when egg-stage failure rates alone were permitted to vary. This most accurately reproduced the full extent of the population decline in the early 1980s. However, reasonable approximates of the decline were obtained whether egg- or chick-stage failure rates or hatching success were allowed to vary in the model.

11.4.2.4Linnet

Siriwardena et al. (2000) modelled the effects of clutch size, brood size, chick:egg ratio and daily nest failure rates on the population size. Breeding performance data was obtained from the BTO nest records database and abundance data from the Common Birds Census. They found that a fall in fledgling production per breeding attempt, driven primarily by increased nest failure rates during the egg period, probably affected abundance, and that this change could have driven the principle population decline (1975-86) for this species.

BTO Nest Record and Ringing Schemes data were used by Moorcroft & Wilson (2000) to model the relationship between breeding productivity (first-egg dates, clutch size, nest survival, partial losses of eggs and nestlings, brood size at fledging) and population size. They concluded that the most likely demographic mechanisms underlying recent population trends are changes in the number of nesting attempts made by females and changes in post-fledging survival rate.

11.4.2.5Red-legged partridge

As for grey partridge, Green (1984) investigated chick diet composition by faecal analysis, identified foraging areas by radio-tracking broods and sampled chick foods by D-vac sampling, sweep netting and ‘quadrat weed samples’. Chick survival rate was measured. The survival of red-legged partridge chicks was positively correlated with the density of Coleoptera and grass spikelets. In conclusion this is a useful dataset as some attempt has been made to sample the invertebrate and plant food items available to foraging birds and a sample size of 14 farms was used.

11.4.2.6Skylark

Chamberlain & Crick (1999) modelled changes in reproductive performance (from Nest Records Scheme data) and declines in the number of breeding skylarks (from CBC data). They found that reproductive performance per nest, in terms of clutch size, brood size and post-hatching survival rate, showed a general improvement between 1975 – 1994, the period in which the number of breeding skylarks declined. They concluded that changes in reproductive performance per nesting attempt were probably not responsible for the decline in numbers, and inferred that number of breeding attempts per pair per season, reductions in the proportion of birds breeding, or increased mortality outside of the breeding season may be responsible (but see comments in Discussion section)..

Siriwardena et al. (2000) modelled breeding performance (BTO nest records data) and population decline (BTO Common Birds Census data) and found no evidence of a relationship.

11.4.2.6Turtle dove

Browne & Aebischer (2001) examined the annual abundance of breeding turtle doves in the UK (BTO data) and compared it to breeding parameters measured in the 1950/60s and in the 1990s (GCT data). The earlier data had been collected at a study site near Newmarket, Cambridgeshire, and the 1990s data was collected by the GCT at study sites in Suffolk and Lincolnshire. Browne & Aebischer’s calculations are quite basic, and the information available in their paper is limited. In conclusion, this dataset is probably not very useful due to the small amount of breeding performance data collected.

Siriwardena et al. (2000) modelled breeding performance (BTO nest records data) and population decline (BTO Common Birds Census data) and found no evidence of a relationship (but see comments in Discussion section).

11.4.2.7Chaffinch, goldfinch, greenfinch, reed bunting, stock dove, tree sparrow and yellowhammer

For these species, Siriwardena et al. (2000) also modelled breeding performance (BTO nest records data) and population decline (BTO Common Birds Census data) and found no evidence of a relationship (but see comments in Discussion section).

11.5Discussion

Few studies have examined relationships between chick food availability and productivity. The only species among those being considered in this report, for which suitable relationships for the modelling approach outlined in section 10 have been published, are two galliforms (grey and red-legged partridges) and one passerine (corn bunting). Studies of other species have been carried out in which both food abundance and productivity have been assessed, but not related to each other e.g. skylark (Poulsen et al., 1998 ); yellowhammer (Stoate et al., 1998; Morris et al., 2002). It is hoped that examples of such relationships may be established from data collected during the ongoing Defra-funded study on “assessing the indirect effects of pesticides on birds” (PN0925).

Measurement of breeding performance is not always straightforward. In the examples modelled in the previous chapter (grey partridge and corn bunting), there was a significant relationship between chick food availability and chick survival. However, absence of such a relationship does not necessarily mean that there is no effect, nor that altering pesticide use may not result in a change in the population. Chick condition may be affected (e.g. Morris et al., 2002), which may influence post-fledging survival. Post-fledging survival has been little studied to date because of the technical difficulties involved. Chick quality is also difficult to measure, and brood reduction (i.e. loss of one or more chick) may affect the performance of remaining chicks, thus masking any measurable effects on individual chick performance. Alternatively, parents may have to work harder to feed their young under conditions of low food availability, which may affect their own condition and hence chances of survival over the following winter.

The ideal measure of productivity is the number of young produced per pair (or per female) per year. Grey partridges and most corn buntings are single-brooded. However, most other species in Table11.1 are multi-brooded. The best available data set measuring breeding output for most species is the British Trust for Ornithology’s Nest Record database. However, this is based on individual nests and does not provide an assessment of annual productivity. Lack of data on total annual productivity does not invalidate the application of the empirically-based modelling approach, but if there is an interaction between outcomes of first and second (or later) broods, this could confuse the interpretation of results based on single nests. For example, failure of the first nest may lead to an early second nest, which may produce young with a better chance of survival than young from a later second nest.

An alternative scenario is that effects of external factors such as pesticides may act differentially on broods produced at different times. For example, Morris et al. (2002) found that yellowhammer foraging density was much lower in fields which had received an insecticide in summer early in the season than later in the season, when unripe grain provided an alternative source of food for nestlings. Chick condition was inversely related to numbers of summer insecticides applied.

Studies relating productivity to population change are available for a greater range of species, however these are not always ideal for incorporation into the modelling approach adopted here. The most wide ranging are the demographic studies undertaken by Siriwardena et al. (2000). These cover a number of years during which populations declined, and for several species including turtle dove, skylark, tree sparrow, yellowhammer, and corn bunting, indicate that breeding performance per attempt was higher while populations declined. However, these models do not allow for density dependence, which could explain this apparently counter-intuitive result. In a declining population, it is likely that sub-optimal habitats will be vacated first, which may improve resource availability on average for the remaining birds, thus potentially increasing productivity. It would be advantageous to carry out further analyses of the BTO’s extensive datasets which incorporate density dependence in the models. However, the modelling approach used in section 10 only estimates the population change from one year to the next, thus approximating to a population in equilibrium and discounting most of the additional factors which influence outcomes over longer periods.