Use of distance sampling to improve estimates of national population sizes for common and widespread breeding birds in the UK

Authors


*Correspondence author. Stuart E. Newson. E-mail: stuart.newson@bto.org

Summary

  • 1Population estimates are of fundamental importance for setting conservation priorities and for numerous aspects of conservation biology.
  • 2Distance sampling, which takes undetected individuals into account, is one of the most widely used methods for generating population estimates. We use this method to generate estimates of the national population size for all common and widespread non-marine breeding birds in the UK using Breeding Bird Survey data.
  • 3There is a strong positive relationship between our distance-sampled estimates and estimates generated using other methods. This implies that most methods get the broad picture right, even if one of them is less precise for individual species.
  • 4For some species, detectability may vary sufficiently between males and females to generate biases in population size estimates if these differences are not taken into account. We found a slight tendency for population estimates from distance sampling to be lower than existing estimates for species with marked sex biases in detectability. This may be a wider problem than is currently acknowledged in distance sampling.
  • 5Distance sampling provides a method for estimating total population size. Other bird population survey methods, such as intensive territory mapping, aim to count the number of breeding pairs, and thus exclude non-breeding individuals. Not surprisingly, we found that distance-sampled estimates tended to be higher for species with a large proportion of non-breeders. Both approaches are valid, but when calculating, reporting and using population estimates attention needs to be paid to which variable is of most interest.
  • 6The new estimates that we present are significantly larger than existing ones for species whose preferred habitat types were not previously well surveyed. This highlights the importance of sampling across all main habitat types.
  • 7Synthesis and applications. This study reviews alternative methods used for producing estimates of population size for common and widespread breeding birds in the UK. We assess a number of factors affecting population estimates generated by distance sampling.

Introduction

Population estimates are of fundamental importance in applied ecology. First, they are frequently used to indicate which species are sufficiently rare that they should be listed as of conservation concern (IUCN 2004). Second, priority- setting frequently uses population estimates to determine if a region holds a significant proportion of a given species’ global population, and thus if the species in question merits special conservation attention, even if its population is not under direct threat in that region (Keller & Bollmann 2004). Third, the design of protected area networks often uses regional population estimates in order to determine which sites are particularly important, within that region, for a given species (Perez-Arteaga et al. 2005). Fourth, habitat-specific population estimates can be used to direct the targeting of conservation attention by identifying which habitat types hold particularly large populations (Gregory & Baillie 1998; Gregory 1999). Finally, population estimates are vital in assessing the ecological determinants of rarity, which may increase extinction risk (IUCN 2004). Such analyses for a limited set of species would enable predictions to be made across a larger sample of species to identify which are at risk of extinction even when their population size is unknown (Kotiaho et al. 2005).

Although the UK avifauna may be the best known in the world, knowledge of national population sizes is far from perfect. Parslow (1973) provided estimates for every species, but in many cases these were little more than informed guesses. Many were improved in the Atlas of Breeding Birds (Sharrock 1976) by using data from the Common Birds Census (CBC) and further improvements were made for the New Atlas (Gibbons, Reid & Chapman 1993). These estimates fell into four classes based on: (i) CBC data; (ii) direct counts of sample areas made during the Atlas fieldwork; (iii) the number of grid squares found to be occupied during the Atlas work; and (iv) species-specific methods. Some of the latter were derived from specifically designed surveys, whilst for other species, data were not available and estimates were based on expert opinion. The estimates published in the New Atlas were subsequently updated by Tucker & Heath (1994), Stone et al. (1997), and Baker et al. (2006), and the figures have also been updated to the year 2000 using the best available information on population trends (BirdLife International 2004).

Both the precision and accuracy of these earlier estimates of national population size are open to debate. In this study, we compare them with new estimates derived from the Breeding Bird Survey (BBS). These new estimates are based on distance-sampling and are derived for all common and widespread non-marine breeding bird species in the UK for which there was a minimum of 60 observations in the BBS data set for the year 2000. This threshold is recommended by Buckland et al. (2001), and here 92 species met this criterion. Whilst a preliminary analysis of BBS data carried out for 20 species has previously suggested that distance-sampling is able to provide useful estimates (Newson et al. 2005), our primary aim here is to review the reliability of the various methods and the overall reliability of the picture that they provide of population size. A secondary aim is to explore the factors promoting differences in national population size estimates between methods in order to improve population estimation in the future. We begin by summarizing other methods used to derive national population estimates, then describe the methods used to calculate BBS-based estimates, and finally derive and test predictions regarding the nature of differences in population estimates generated by the different methods.

Methods

existing population estimates

These data were derived using four different methods. For the majority of species (51 of 92), territory mapping data from 144 CBC plots surveyed in 1989 were the basis (Marchant et al. 1990). Estimates of density were generated for three habitat types: woodland, farmland and ‘other habitats’. Densities in the first two habitat types were reported by the CBC; estimates for other habitats were the average of woodland and farmland densities weighted by the proportion of these habitats occupied. Estimates were corrected for geographical bias, using data on the number of 2 × 2 km squares (tetrads) occupied per 10 × 10 km square obtained from the second breeding bird atlas (Gibbons et al. 1993).

For 18 species, population estimates were derived from the number of 10 × 10 km squares occupied during surveys for the second breeding bird atlas. Data are based on two 1-h counts during early and late parts of the breeding season, to at least eight tetrads within each 10 × 10 km plot and supplemented with opportunistic records. The number of occupied 10 × 10 km squares was multiplied by an informed estimate of pairs per occupied square.

For six species, counts recorded during the second breeding bird atlas were used to estimate population density, which was multiplied by the area of the UK to obtain the population estimate. As tetrads were not selected randomly, population densities from randomly located point counts in approximately 15% of 10 × 10 km squares were used to correct for variation in densities in tetrads that were and were not visited.

For the remaining 17 species, estimates were obtained from species-specific information. Some were derived from targeted national surveys (e.g. mute swan Cygnus olor), which we believe to be of good quality. Others were essentially guesswork (e.g. feral pigeon Columba livia).

bbs-based population estimates

For all comparisons, new estimates were derived from BBS data collected in 2000 to be comparable with estimates in Birdlife International (2004). Because of its applied value, we also present population estimates derived from BBS data using the same methods for 2006, the most recent year for which data were available at the time of writing this study. The BBS is carried out annually on a stratified random sample of over 2500 1-km squares. BBS squares are stratified regionally by human population density in order to make the most of volunteer resources. These regions are referred to as sampling regions hereafter. Fieldwork involves two visits to each square, the first between early April and mid-May and the second between mid-May and late June. Birds are recorded in each 200-m section along two 1-km transects. Their perpendicular distances from the transect line are recorded in three distance bands (0–25 m, 25–100 m, 100 m or more). The number of distance bands is a compromise between collecting useful data and the ease with which volunteers can carry out the survey. However, even with a few bands, resulting population estimates will be unbiased provided that birds are allocated to the correct band (Buckland et al. 2001). The number of BBS squares recording each species, number of observations of each species, and the percentage of observations in the 0–25 m band in 2000 are shown in Table S1 (Supplementary material). Flying birds actively using resources in a square, such as displaying skylarks Alauda arvensis, are assigned to the appropriate distance band; records of flying birds are excluded from analyses.

Observers exclude juvenile birds from counts where possible because their inclusion would inflate estimates of the breeding population. Such errors are more likely for species that breed early, mainly residents, and those counted during the late visit. Conversely, most long-distance migrants are still arriving in the UK during the first visit period; thus, counts of these species during this visit may underestimate densities. To control for these biases, we only use data from early visits for resident species and only data from late visits for migrants. The exception is corn bunting Emberiza calandra, a late-breeding resident, for which we used data from the late BBS visit only (May to late June).

For the colonial nesting rook Corvus frugilegus and sand martin Riparia riparia, counts of active nests are also recorded. For these species, we use the maximum nest count multiplied by 2 to be equivalent to the number of adults, where this value exceeds the count of individual birds.

The main habitat within each 200-m transect section was recorded by observers according to Crick (1992).

Detectability must be taken into account when converting counts into densities, and subsequently, population estimates. This was achieved using distance sampling software (distance 5·0; Thomas et al. 2006). Birds recorded in the final distance band (100 m or more) were excluded from analyses, because counts within an unbounded category are difficult to interpret.

Whether birds are in flocks is not recorded in the BBS, and the majority of British species tend not to form flocks during the breeding season. Therefore, we assume that counts are of individual birds for analyses. Even for species that sometimes flock during the breeding season, such as the barn swallow Hirundo rustica, this approach is likely to yield a reliable population density estimate if these four conditions are met: (i) all flocks on the transect line are detected; (ii) mean flock size is not very large; (iii) flock size is not highly variable; and (iv) detectability of flocks is not very low (Buckland et al. 2001). For species in this study, with the exception perhaps of common linnet Carduelis cannabina, these conditions appear likely to be met.

Habitats differ in their structure which is likely to influence detectability. Detectability may also vary between regions for a variety of reasons. The structure of a single habitat may differ with geographical location (e.g. denser woodlands in southern UK), for example. Population densities may differ between regions, and in areas with high densities, individual birds may need to invest more time in territory defence, singing more frequently and increasing their detectability. We took heterogeneity in detectability arising from variation between habitats and regions into account by including these as factor covariates in a global multiple-covariate detection function (Marques & Buckland 2003). This approach uses the whole data set to provide information about the shape of the detection function, so it was not necessary to fit a separate detection function to each habitat–region combination.

Following Buckland et al. (2004), we fitted half-normal and hazard-rate distributions to the data. We identified nine main habitat types (broad-leaved, coniferous and mixed woodland, semi-natural grassland, heath and bog, arable, pastoral and mixed farmland, human habitats and water bodies), and 12 regions (nine English Government Office Regions, Wales, Scotland and Northern Ireland) and adopted the following stepwise approach. For each species, we estimated f(0), i.e. the value of the probability density function of perpendicular distances at zero distance, without and including one and then both habitat or region as covariates to both half-normal and hazard rate models. Akaike's Information Criterion (AIC) was used to identify the best-fitting model for each species. This model was then applied to the encounters from surveyed squares to produce an estimate of individuals within each square. Whilst habitat was retained as a covariate for all species, for a quarter of species (23 of 92), region was also retained. This suggests that even when habitat effects are taken into account, detection functions may vary spatially.

To obtain estimates of population size and confidence intervals, we used a bootstrap resampling procedure of 399 iterations (Crowley 1992). For each iteration, 1 km squares were randomly resampled with replacement within the sampling region, the sample drawn being equal to the total number of 1 km squares in that region. The values for each sampling region were summed to derive an estimate of each species in that region and region totals summed to give an estimate of total population. The 10th and 390th ordered bootstrap values across iterations and total population estimate were taken to give the lower and upper 95% confidence limits of the national estimates, respectively.

Most estimates presented in BirdLife International (2004) were of breeding pairs. For comparative purposes, we multiplied these by 2 to get estimates of individual birds. Statistical comparisons in this study are based on log10-transformed population size estimates.

expectations

Overall effects

We expected BBS-based estimates to be positively correlated with previous estimates, and the relationship to be monotonic. However, we had no general expectation as to the strength of the correlation or exact form of the relationship.

Habitat effects

We categorized each species according to these five broad habitat categories: (i) woodland; (ii) human habitats (urban, suburban and rural); (iii) farmland (arable and pastoral); (iv) semi-natural grassland, heath and bog; and (v) wetlands (standing and flowing water). We used a smaller number of habitat types than that used to model habitat differences in detectability because (i) habitat types were maximized in detectability modelling to increase precision in taking detectability differences into account, and (ii) the four woodland categories used to model detectability were merged, as our primary aim was to distinguish between species whose characteristic habitat types corresponded to those covered well by the CBC (i.e. woodland and farmland). We took the proportion of BBS 200-m sections in each class as the availability of that class (p) and the proportion of sections on which a species occurred that were in that class as the species usage of that class (r). We calculated Jacobs (1974) preference index of the species for each habitat:

J = (r – p)/[(r + p) – 2rp]

The habitat with the largest index for that species was taken to be its characteristic habitat, see Table S1.

Previous population estimates for 51 common farmland and woodland species were based on CBC territory mapping and are likely to be fairly accurate. The BBS-based population estimates for such species are also likely to be unaffected by marked bias, and we thus expect the two sets of estimates to be similar. Conversley, previous estimates for species whose main habitat type is neither farmland nor woodland are less likely to be accurate. For these species, estimates are based on population densities in farmland and woodland, which are not the focal species’ preferred habitat, and this may lead to their national population size being underestimated. For such species, we expect BBS-based estimates to be higher than previous estimates.

Methodological effects

Estimates derived from CBC data are based on the number of territories, whilst BBS-based estimates concern the number of adult birds. For all species, there are likely to be non-territory holders that will not be recorded by the CBC (Newton 1998). This is likely to be greatest for species which delay breeding until their second year or older, and thus have a large proportion of non-breeding individuals in their population. Therefore, we expect that BBS estimates will be larger than previous estimates for species which delay breeding. Using the ornithological literature (Cramp, Simmons & Perrins 1977–1994), we classified species as those in which the majority of adults start breeding from 2 or more years of age, 3 or more years, and where the majority of individuals start breeding in their first year (see Table S1).

Counts made during timed tetrad visits may have missed some birds, whilst estimates derived from BBS data explicitly use distance sampling to take undetected individuals into account, giving higher estimates.

Whilst previous estimates based on the number of occupied squares are of lower quality than those derived from CBC or count data (Stone et al. 1997), we did not expect the former, or previous estimates based on species-specific methods, to differ from BBS-based estimates in any systematic way.

Geographical bias

The CBC was geographically biased towards southern England. Within a rectangle of easting 3000 and northing 5000 of the National Grid, which we refer to as ‘Southern UK’, the CBC farmland plots were representative of the general landscape (Fuller, Marchant & Morgan 1985). Because we have produced estimates of population size for sampling regions, we use these to estimate the proportion of the UK populations of each species lying within and outside Southern UK. If corrections for geographical bias applied to the CBC data were under- or overcompensated for, UK population estimates for species predominantly outside Southern UK would be systematically biased when based on BBS data. To test this, we examined two groups: species for which > 60% and 70% of the population lie outside Southern UK (see Table S1). The latter group is more likely to be biased, but the former group provides a larger sample.

Effects of sex on detectability

The BBS does not ask observers to record sexes separately even if they are identifiable by plumage or through differences in behaviour (e.g. singing/displaying males). For species where there is a significant difference in detectability between the sexes, analysing the data for the two sexes combined may underestimate true population size. Simulations carried out by the authors to examine the influence of such differences in detectability and work by Buckland et al. (2001) have shown that distance sampling is robust to fairly large sex-based differences in detectability. However, this breaks down at levels where detection is mainly of one sex. For this reason, using the ornithological literature and an independent assessment by experienced field ornithologists, we focus on species for which sex bias in detectability is likely to be extreme (see Table S1). We expected that BBS-based population estimates would be smaller for species with a marked difference in detectability between sexes.

Statistical analyses

We used paired t-tests to examine whether BBS and previous estimates were significantly different from one another for species associated with each of the five habitat classes. We used Generalized Linear Models to test for the effects of previous survey methods, delayed breeding, geographical bias and influence of sex on detectability on the difference between BBS and previous estimates, defined as log10(BBS/previous). Multiple regression analysis was used to quantify the extent to which habitat, previous survey method, delayed breeding, geographical bias and influence of sex on detectability explain differences between BBS and previous estimates. For these analyses, an information theoretic approach to model selection was used (Burnham & Anderson 2002). All analyses were carried out in sas (SAS Institute 2001).

Results

overall effects

BBS-based population estimates are broadly comparable to previous estimates (Table 1, Fig. 1). Within the three orders of magnitude covered by the 92 species in this study, there was a tendency for BBS-based estimates to be higher than previous estimates for less abundant species (e.g. ratio of logged BBS to previous estimates is 1:0·9 when the former = 4·5, but was 1:1 when logged BBS estimate = 7·5). Across orders of magnitude, the slope of the fitted regression (0·928; principal axis regression) did not depart significantly from unity (95% CI of slope: 0·855–1·006).

Table 1.  Estimates of UK population size for breeding birds calculated from the Breeding Bird Survey (BBS) in 2000 and previous estimates corrected to the year 2000 (BirdLife International 2004). Because of their immediate applied value, we also present BBS-based estimates calculated for 2006. Ninety-five per cent confidence intervals are presented for BBS-based estimates. For the previous estimates, we present the minima and maxima when these were provided in the original sources, with the exception of Canada goose for which the original source provided 95% confidence intervals. Figures are rounded to the nearest 1000 individuals. Species are ordered alphabetically within previous survey method. Where analyses suggests that the 2000 BBS or previous estimates are most likely to represent real population levels, we highlight these in bold
SpeciesThis study (2000)Previous estimatesThis study (2006)
Thousands of birdsThousands of pairs × 2Thousands of birds
Common Birds Census
Red-legged partridge (RL)485391–579272144–400496438–555
Grey partridge (P.)175128–222145140–150187145–229
Moorhen, common (MH)369312–426540 355312–399
Stock pigeon (SD)280225–335618 241204–278
Wood pigeon, common (WP)84187811–902457305140–632010 0389450–10 627
Collared dove, Eurasian (CD)16741484–1863596 20451846–2245
Turtle dove, European (TD)7961–9788 3728–46
Skylark (S.)24702145–27943570 21461913–2379
Swallow, barn (SL)16771441–19121452 23142012–2617
Meadow pipit (MP)64085476–73403360 45503967–5134
Tree pipit (TP)172119–226149 182114–251
Pied wagtail (PW)999850–1148624544–7041032902–1162
Yellow wagtail (YW)12794–1603823–539976–121
Wren, winter (WR)96849109–10 25917 024 82017788–8615
Hedge accentor (D.)37933519–40674326 44134147–4679
Robin, European (R.)85718071–907111 790 82877876–8697
Redstart, common (RT)242191–294202 191138–244
Song thrush (ST)15531405–17022288 13831270–1497
Mistle thrush (M.)544466–622445 433376–489
Blackbird, common (B.)10 2549658–10 8509870 10 0269562–10 489
Garden warbler (GW)268226–310380 219185–254
Blackcap (BC)12321125–13401864 14621351–1572
Lesser whitethroat (LW)9170–111128 10285–120
Whitethroat, common (WH)19211715–21271890 22362056–2416
Sedge warbler (SW)735551–919642 514393–636
Willow warbler (WW)32602916–36044250 25932313–2872
Chiffchaff, common (CC)608547–6691614 716657–775
Goldcrest (GC)17531441–20661684 13081127–1488
Spotted flycatcher (SF)278224–332127 308226–391
Great tit (GT)39643712–42164148 57745472–6076
Coal tit (CT)993812–11751306 12621072–1453
Blue tit (BT)88118288–93347070 11 34610 782–11 910
Marsh tit (MT)8258–106105 7356–90
Willow tit (WT)219–3417 2313–34
Long-tailed tit (LT)998881–1115546 1043931–1156
Nuthatch, wood (NH)153127–179288 201175–228
Treecreeper, Eurasian (TC)212164–260428 319253–385
Magpie, black-billed (MG)13221202–14421300 11511063–1238
Jay, Eurasian (J.)194160–227320 210175–246
Jackdaw, Eurasian (JD)23152042–25891110 25182221–2816
Carrion crow (C.)18551673–20372404 19971826–2168
Starling, common (SG)65995719–74791608 49854398–5573
Tree sparrow, Eurasian (TS)365262–468136 429313–545
Chaffinch (CH)14 06613 242–14 89011 948 14 34713 709–14 986
Linnet, common (LI)23071964–26491112 19541704–2203
Goldfinch, European (GO)18871683–2092626 19351736–2134
Greenfinch, European (GR)38913482–43001468 43704084–4657
Bullfinch, common (BF)371292–450332 397324–471
Reed bunting (RB)410324–496403384–422594480–707
Yellowhammer (Y.)25992306–28931584 24532234–2672
Corn bunting (CB)9367–1182117–249773–120
Atlas counts
Great crested grebe (GG)239–3612 2113–30
Greylag goose (GJ)13464–2043532–3810051–149
Shelduck, common (SU)7143–991712–226029–91
Coot, common (CO)222170–2745145–58235184–287
Sand martin (SM)15365–242275133–42224379–406
Raven, common (RN)173–3126 4317–69
Number of occupied squares
Kestrel, common (K.)6349–7874 5040–61
Pheasant, common (PH)12351116–135437003600–380015341415–1652
Common sandpiper (CS)9843–15424 7030–111
Cuckoo, common (CK)3118–453019–404022–58
Tawny owl (TO)2911–4839 74–10
Little owl (LO)1811–251712–23139–18
Swift, common (SI)227157–29712040–200195142–248
Green woodpecker (G.)7564–8748 10493–115
Great spotted woodpecker (GS)166142–1918174–89250228–272
House martin (HM)905708–1101808546–10701220860–1580
Grey wagtail (GL)13093–1688577–9211173–149
Wheatear, northern (W.)578277–880105 399280–518
Whinchat (WC)8335–1323322–4412449–198
Stonechat (SC)14085–1946939–99201133–269
Grasshopper warbler, common (GH)3116–4525 176–28
Reed warbler, Eurasian (RW)219156–282183122–244257204–311
Lesser redpoll (LR)16397–22954 13568–202
Siskin, Eurasian (SK)353192–514738 348215–482
Single species surveys
Grey heron (H.)5840–7530 7245–100
Mute swan (MS)10473–1355049–51181133–230
Canada goose (CG)179122–2359390–97223174–273
Mallard (MA)14311191–1671222126–31816181393–1843
Tufted duck (TU)8453–1152220–23188141–235
Buzzard, common (BZ)10780–1337562–88130101–160
Sparrowhawk, Eurasian (SH)2918–3980 4834–61
Willow ptarmigan (RG)296162–430310 168110–225
Oystercatcher, European (OC)379227–532226197–254302206–398
Golden plover, European (GP)9234–1519877–1194323–63
Lapwing, Northern (L.)507365–649311274–348318252–384
Redshank, common (RK)8216–1487663–894824–71
Curlew, Eurasian (CU)135102–169225199–250155119–190
Snipe, common (SN)230116–343122105–13814098–182
Feral pigeon (FP)14321153–1711350200–5001178960–1396
Rook (RO)21521837–246725502240–286021241766–2482
House sparrow (HS)10 0808883–11 27857754200–735010 4399489–11 389
Figure 1.

Comparison between BBS-based estimates of population size for breeding birds in the UK and previous estimates for 2000 (BirdLife International 2004). Species codes are given in Table 1. The solid line is the line of equivalence, while the broken lines show where BBS estimates are twice previous estimates and vice versa.

The lack of confidence intervals for previous population estimates prevents a formal statistical comparison of the number of species for which the two population estimates are significantly different. However, there are clearly large differences for some species (Fig. 1), with 23 species (25% of total) having BBS-based estimates that are at least double previous estimates (e.g. mallard Anas platyrhynchos; feral pigeon Columba livia; northern wheatear Oenanthe oenanthe; and common starling Sturnus vulgaris) and six species (7% of total species) having BBS-based estimates that are half those of previous estimates (Eurasian sparrowhawk Accipter nisus; common pheasant Phasianus colchicus; stock pigeon Columba oenas; treecreeper Certhia familiaris; common chiffchaff Phylloscopus collybita; Eurasian siskin Carduelis spinus).

habitat effects

As expected, BBS-based estimates were significantly higher than previous estimates for species associated with human habitats (on average 2·0 times as high, t16 = 4·73, P ≤ 0·001) and wetlands (on average 2·2 times as high, t21 = 3·94, P = 0·001), whilst estimates were not significantly different for woodland species (t27 = 1·15, ns) or farmland species (t16 = 1·04, ns). Contrary to expectations, there was no evidence that BBS-based estimates for species associated with semi-natural grasslands, heath and bogs were higher than previous estimates (t8 = 1·69, ns), although the number of species in this group and hence power to detect a diference is small.

methodological effects

BBS-based and previous population estimates did not differ as a function of the method used (F3,88 = 2·59, ns).

BBS-based estimates tended to be higher than previous estimates for species that delay breeding until 3 or more years of age (on average three times as high, F1,90 = 4·50, P = 0·04), although not for species breeding from 2 or more years of age (F1,90 = 0·15, ns). This supports the expectation that BBS-based estimates based on the number of adults tend to be higher than previous estimates based on breeding adults, for species for which there is likely to be a sizeable non-breeding population.

geographicAL bias

For estimates based on CBC data, BBS-based and previous population estimates did not differ according to whether the species’ distribution was centred on the Southern UK (60% or more of population outside Southern UK: F1,49 = 0·67 ns; 70% or more outside Southern UK: F1,49 = 0·08 ns). This suggests that corrections for geographical bias applied to previous estimates were generally sufficient.

effects of sex on detectability

BBS-based and previous population estimates did differ as a function of sex difference in detectability (F1,90 = 6·41, P = 0·01), with BBS-based estimates being smaller (on average 8% lower) in species with a marked difference in detectability. This supports the expectation that the BBS underestimates population size for species for which there is a large difference in detectability between sexes.

multiple regression

After adjusting for the significant effects of habitat, delayed breeding and sex-specific detectability by regressing the log (BBS/previous estimate) against these variables, the discrepancy between BBS and previous estimates is reduced, although there is still much unexplained variance as shown by the low r2 values for the models in the 95% confidence set (Table 2, most parsimonious model r2 = 18·0%). Whilst the 95% confidence set of models contains a large number of alternative models, all of these retain habitat type as a predictor, and this variable consistently has the highest partial r2 values (model averaged partial r2 = 13·1%). Delayed breeding, sex-biased detectability and geographical bias are each retained in four models. However, the respective model averaged partial r2 values for these variables (1·3%, 0·9% and < 0·1%) are low.

Table 2.  Multiple regression models of the difference between BBS-based population estimates and previous estimates. For each predictor retained in the model, we provide its partial r2. Model selection followed an information theoretic approach. Values for Δi express the difference in AIC between each model and the most parsimonious model, i.e. that with the smallest AIC value. Model weights indicate the probability that each model provides the best fit to the data. We present the 95% candidate set of models, i.e. those whose cumulative model weights sum to 0·95
Habitat (%)Delayed breeding (%)Sex-biased detectability (%)Geographical bias (%)Survey (%)ΔiModel weightModel r2
13·12·82·0  0·00·2970·180
14·11·8   1·00·1800·160
14·1 1·1  1·70·1270·152
14·1    2·00·1040·141
14·3 0·9< 0·1 2·60·0770·152
14·1    0·1 3·00·0630·143
14·31·7 < 0·1 3·70·0470·160
13·32·70·2< 0·1 4·70·0280·180
13·6   0·64·90·0260·142

Discussion

Because we have no baseline knowledge of actual population sizes for most widespread species in the UK, it is difficult to assess whether BBS-based population estimates are more accurate than their predecessors. Whilst a comparison of independently derived population estimates provides insight into the bias inherent in different survey methods, which influences our confidence in such estimates, the best solution may be to carefully design independent surveys for carefully chosen species that would provide the most reliable baseline for comparison.

Broadly speaking, in this study the previous national population estimates and the new estimates derived from BBS data are consistent. Overall, there is a strong relationship between the two but there are also some important differences.

Population estimates derived from the BBS tend to be over twice as large as previous estimates for species that prefer human habitats (e.g. collared dove Streptopelia decaocto) or wetlands (e.g. mallard Anas platyrhynchos); in contrast, the BBS and previous population estimates are generally similar for woodland and farmland species. This agrees with previous suggestions that the population size of some species with large concentrations in urban areas had previously been underestimated (Bland, Tully & Greenwood 2004; Newson et al. 2005) because they were largely derived from estimates of their population densities in less-preferred habitat types (woodland and farmland). In contrast, the BBS samples across all major habitat types yielded more accurate estimates for species associated with habitats other than woodland and farmland.

The larger population estimates derived from the BBS for wetland species may partly arise because many wildfowl species have a large non-breeding component to their population (Cramp et al. 1977–1994; Owen et al. 1986). This also occurs in many raptor species, such as the common buzzard Buteo buteo, for which there are approximately three non-breeding individuals for every paired bird (Kenward et al. 2000). The discrepancy between BBS and previous estimates for the common buzzard is not as great as expected, but there were still almost one-and–a-half times as many common buzzards calculated by the BBS than by previous methods. We find evidence that species that typically do not breed until they are 3 years old tend to have higher population estimates when these are calculated using distance sampling compared to previous methods. Once we control for habitat, the additional variance explained by delayed breeding, whilst small, suggests that the difference between estimates is in part because most previous methods, such as the CBC, aim to count only breeding pairs rather than all adults. In general, population estimates should explicitly state whether such non-breeding individuals are included, and what proportion of the estimate they constitute. Non-breeding individuals will often commence breeding if existing territories become vacant, or suitable breeding habitat is created, and thus have the potential to contribute to population growth. They also clearly influence resources used and contribute to assemblage structure. In much of macroecology, the total population size is more relevant than the breeding population size, although it is less often available. Our analyses highlight the importance of clearly defining what is meant by the population of a given species, particularly during the breeding season (Greenwood, Crick & Bainbridge 2003).

For species for which there is thought to be a significant sex difference in detectability, there is a tendency for BBS-based estimates to be smaller than previous estimates. After controlling for habitat, the additional variance explained by sex difference in detectability, whilst small, suggests that modelling a single detection function for these species, based predominantly on detections of singing males, is likely to underestimate the real population size. For these species, previous estimates are likely to be more accurate than BBS-based estimates. Biases generated by sex differences in detectability are likely to apply to distance sampling of taxa other than birds (e.g. Schmidt 2004), and users of distance sampling need to pay attention to such biases.

The BBS-based national population estimates of some species of conservation concern, such as spotted flycatcher Muscicapa striata, are much larger than current estimates. In the UK, the assignment of conservation priorities of relatively abundant species is based on changes in the rate of population decline (Gregory et al. 2002). Our results do not indicate that conservation status needs to be re-assessed, but that many more individuals of such species have been lost than previously realized.

Our analyses demonstrate that national population estimates derived from distance sampling are qualitatively similar to previous estimates. For a third of species, however, important quantitative differences exist. These differences highlight a number of important effects. First, albeit not surprisingly, sampling across all major habitat types yields different, and for some species probably superior estimates to those calculated from sampling a sub-set of main habitat types. Secondly, a difference in detectability between males and females of the same species may generate marked biases in population size estimates; yet, such differences are rarely taken into account. Thirdly, an important but often neglected source of variation in detection functions may be that between geographical regions. Finally, and more generally, when calculating, reporting and using population estimates, it is important to consider, and state explicitly, whether the unit of interest is total population size, or breeding population. By highlighting these issues in relation to UK bird monitoring data, we facilitate the production of more precise estimates of population size for birds and other taxa where detectability is less than perfect.

Acknowledgements

We would like to thank all the volunteer fieldworkers. The BBS is jointly funded by the BTO, JNCC and RSPB. K.L.E. was supported by The Leverhulme Trust and NERC, and K.J.G. holds a Royal Society-Wolfson Research Merit Award. We thank Steve Freeman and Rob Robinson for analytical advice and Jeremy Wilson, Nicholas Aebischer, Richard Gregory and Frederic Jiguet improved an earlier draft.

Ancillary