Adverse effects of agricultural intensification and climate change on breeding habitat quality of Black-tailed Godwits Limosa l. limosa in the Netherlands


Corresponding author.


Agricultural intensification is one of the main drivers of farmland bird declines, but effects on birds may be confounded with those of climate change. Here we examine the effects of intensification and climate change on a grassland breeding wader, the Black-tailed Godwit Limosa l. limosa, in the Netherlands. Population decline has been linked to poor chick survival which, in turn, has been linked to available foraging habitat. Foraging habitat of the nidifugous chicks consists of uncut grasslands that provide cover and arthropod prey. Conservation measures such as agri-environment schemes aim to increase the availability of chick foraging habitat but have not yet been successful in halting the decline. Field observations show that since the early 1980s, farmers advanced their first seasonal mowing or grazing date by 15 days, whereas Godwits did not advance their hatching date. Ringing data indicate that between 1945 and 1975 hatching dates advanced by about 2 weeks in parallel with the advancement of median mowing dates. Surprisingly, temperature sums at median mowing and hatching dates suggest that while the agricultural advancement before 1980 was largely due to agricultural intensification, after 1980 it was largely due to climate change. Examining arthropod abundance in a range of differently managed grasslands revealed that chick food abundance was little affected but that food accessibility in intensively used tall swards may be problematic for chicks. Our results suggest that, compared with 25 years ago, nowadays (1) a much higher proportion of clutches and chicks are exposed to agricultural activities, (2) there is little foraging habitat left when chicks hatch and (3) because of climate change, the vegetation in the remaining foraging habitat is taller and denser and therefore of lower quality. This indicates that for agri-environment schemes to make a difference, they should not only be implemented in a larger percentage of the breeding area than the current maxima of 20–30% but they should also include measures that create more open, accessible swards.

Agricultural intensification is one of the main drivers of the decline of European farmland birds (Donald et al. 2001, 2006). For some species, adult survival has been adversely affected by changing cropping patterns, increasing agro-chemical use, increasing stocking rates and other changes associated with less labour-intensive, more industrial forms of agriculture (Siriwardena et al. 2000, Peach et al. 2003, Newton 2004). Many other species have suffered declines in their reproductive success because of the shift towards more high-input forms of agriculture (Newton 2004, Schekkerman et al. 2008, 2009).

How changing farming practices affect the quality of the breeding habitat has been well-examined for a number of arable species in the UK, such as Grey Partridge Perdix perdix (Potts & Aebischer 2008) and Skylark Alauda arvensis (Donald et al. 2002). Less information is available on how the quality of the breeding habitat of grassland species has been affected by changing farming practices (McCracken & Tallowin 2004).

Arguably, the effects of agricultural intensification in grassland areas can be categorized by two types of changes. First, intensification will change the timing of agricultural activities, generally moving them forward and enabling them to occur in a more rapid succession. This gives ground-nesting birds less opportunity safely to hatch or fledge their chicks on agricultural land. Effects of agricultural intensification on the timing of agricultural activities are, however, confounded with the effects of climate change (e.g. Both et al. 2005). The last few decades have seen a general increase in winter and spring temperatures resulting in a more rapid vegetation development, which could also explain an advance of the first mowing or grazing date. Secondly, agricultural intensification affects the physical and biotic characteristics of the fields. Increased drainage, the removal of relief, re-seeding and fertilization create dense, homogeneous swards and less penetrable soils that are less suitable for breeding or foraging (McCracken & Tallowin 2004).

Here we assess how agricultural change in the Netherlands has affected the quality of the breeding habitat of Black-tailed Godwits Limosa l. limosa, and how this is influenced by climate change. The Black-tailed Godwit, a grassland breeding wader, is one of the few species for which the Netherlands holds an international conservation responsibility. Approximately 40% of the European Godwit population breeds in the Netherlands (Burfield et al. 2005). As a result, the species has been studied extensively and its ecology, population dynamics and the effects of various conservation measures are much better known than that of other Dutch farmland birds (e.g. Beintema & Visser 1989, Beintema et al. 1991, Kleijn et al. 2001, Gill et al. 2007, Schekkerman & Beintema 2007, Schekkerman et al. 2008, 2009). Approximately 85% of the Black-tailed Godwits in the Netherlands breed on agricultural grasslands (as opposed to grassland reserves). On farmland, the first agri-environment scheme targeted at this and other grassland breeding birds started in 1981. In recent years, approximately €34 million was spent annually on nationwide measures to enhance grassland breeding birds in general and the Black-tailed Godwit specifically (e.g. Van Brederode & Laporte 2006, Verhulst et al. 2007, Schekkerman et al. 2008) with 89% of that budget allocated to agri-environment schemes. Nevertheless, the Dutch breeding population has declined at about 3% per year since 1990 (and with almost 6% in the last 5 years) and in 2004 numbered around 62 000 compared with 85 000–100 000 around 1990 (Osieck & Hustings 1994, Teunissen & Soldaat 2006).

Dutch agricultural fields are among the most intensively farmed lands in the world (Giampietro et al. 1999). Grasslands in good Black-tailed Godwit breeding areas receive on average approximately 250 kg N/ha/yr (Kleijn et al. 2001, Verhulst et al. 2007) and the crop may be removed by cutting or grazing as often as 5–10 times per year. Godwit chick survival is the main bottleneck for population growth (Schekkerman & Müskens 2000, Schekkerman et al. 2008). The nidifugous Godwit chicks forage in 15-to 30-cm-tall, preferably uncut, swards from which they glean arthropods and that provide cover against predators (Schekkerman & Beintema 2007). Godwit chicks eat a wide range of invertebrates (Beintema et al. 1991); Diptera, Coleoptera and Hymenoptera were the three most frequently encountered arthropod groups in Godwit chick faecal samples (Beintema et al. 1991) and these groups are generally the most abundant arthropod groups in Dutch agricultural wet grasslands (Schekkerman & Beintema 2007). It is unknown whether chicks select specific arthropod groups or whether prey uptake is supply-driven. Chick survival is positively related to the area of uncut swards, but even in areas where the supply of uncut vegetation has been enhanced by agri-environment schemes, chick survival is generally too low to compensate for adult mortality (Schekkerman et al. 2008). It is unknown whether this is caused by an insufficient quantity of foraging habitat, by the fact the uncut swards are of too poor a quality, or both.

In this study we examine temporal changes in the agricultural activities most relevant to Godwit reproduction, assess whether this is related to climate change and determine how this interferes with the laying and hatching period of Black-tailed Godwits. Furthermore, we examine how farming intensity is affecting the quality (food availability) of the foraging habitat of Godwit chicks. Finally, we discuss the implications of our findings for conservation management aimed at Black-tailed Godwits in agricultural grasslands.


Changes in timing of agricultural activities and breeding Godwits

To assess whether changes in the timing of agricultural activities have resulted in an increased exposure of Black-tailed Godwit clutches and chicks to agricultural activities, we reanalysed data from Kruk et al. (1996), Schekkerman (1997), Schekkerman and Müskens (2000) and Schekkerman et al. (1998, 2005). All studies presented data for a number of different areas and spring dates, on the number of fields and the proportional area which had been cut or grazed for the first time during the growing season. For each of these study areas we obtained the date at which 50% of the area had been grazed or mown for the first time (henceforth called ‘median mowing date’; 33 year–area combinations). For 24 year–area combinations we also obtained the median hatching dates of Godwit chicks, derived from direct observation, predictions by egg flotation (Van Paassen et al. 1984) and back-calculation from chick biometrics.

The sum of the positive mean daily temperatures since 1 January (temperature sum or Tsum) determines both the start of the growing season and the subsequent vegetation growth rate (e.g. Prins et al. 1988). The first seasonal mowing date and the median laying and hatching dates of grassland breeding waders are related to the temperature sum (Kruk et al. 1996). Temperature sum is therefore a good indicator of the effects of climatic changes on both the timing of agricultural activities and that of incubating birds. Daily mean temperature data were obtained from the KNMI (Royal Netherlands Meteorological Institute; For each area we used data from the nearest weather station (average distance 16 km, se 1.5 km).

We first examined the trends in median mowing and hatching dates between 1982 and 2005 using ordinary least-squares regression analysis. We subsequently examined the trends in the temperature sums at the median mowing and hatching dates between 1982 and 2005. We assumed that significant relationships between temperature sums and time were indicative of climate change, whereas significant (additional) relationships between mowing date or Godwit hatching dates and Julian date after accounting for effects of Tsums were indicative of effects of changes in agricultural practices.

To explore changes in Godwit breeding phenology in relation to changes in climate and agricultural practice at a larger spatial and temporal scale, we analysed the average dates on which Godwit chicks have been ringed in the Netherlands, as reported by Beintema et al. (1985) for 1911–1970, and extended with data stored in the database of the Dutch Centre for Avian Migration and Demography. Prior to 1958, median ringing dates were calculated over periods of 6–12 years to obtain adequate sample sizes; for 1958–2007 annual medians were calculated. To estimate hatching dates, the median date reported by Beintema (1995) for 1976–1985 was subtracted from the median ringing date in the same years to obtain the average age at ringing (13 days), which was assumed to have remained constant over time until 2002. In 2003–2006 intensive radiotelemetry studies (Schekkerman et al. 2008, 2009) sought to ring chicks directly after hatching, and average age at ringing was 4 days, which was used in the calculation for these years. Other assumptions that must be made for ringing dates to reflect accurately changes in laying and hatching dates are that any date-dependency of mortality of nests or young chicks and the propensity of Godwits to renest after clutch loss did not change over time.

The development of winter and spring climate during the last century was quantified as the Tsums accumulated at De Bilt, central Netherlands, at the end of each month from January to June, and on the annual median hatching dates of Godwit chicks. To illustrate trends, smoothing splines (with 5 degrees of freedom) were fitted to the annual (or periodical) median hatching dates and Tsum dates by means of generalized additive models (SSPLINE directive in genstat 8.11; Payne et al. 2002). The degrees of freedom, and therefore the flexibility of the spline, was selected by estimating visually how well the curvilinear relationship fitted the general trend in the data.

Estimated changes in the timing in the first mowing date were taken from Beintema et al. (1985) for the period 1910–1970 and from Teunissen et al. (2008) (24 site–year combinations) for the period 1997–2005.

The relation between land-use intensity and quality of chick foraging habitat

The quality of Godwit chick foraging habitat is determined by the amount of food available in it, the accessibility of that food and the level of protection it provides against predators. The relative importance of these factors, whether this changes with chick age and how they are interrelated are unknown. Here we use food availability (the abundance, biomass and mean body size of arthropods) as a proxy for the quality of chick foraging habitat. Similarly, land-use intensity is characterized by many different factors such as yield, fertilizer use, stocking density or level of mechanization (Kleijn et al. 2009b). In western European grasslands these activities are generally aimed at producing as much crop in as short a time as possible. We therefore estimated land-use intensity as the vegetation height on 16 May (average median hatching date of Godwit chicks in the last decade) and fertilizer application rate.

In 2006, arthropods were sampled in wet grasslands in two agricultural areas with high densities of Black-tailed Godwits, the Gerkesklooster area (= 6; 53°14′36″N, 6°10′11″E) and the Ronde Hoep area (n = 6; 52°16′07″N, 4°54′00″E). In 2007, 16 fields were sampled in the meadow bird reserve ‘Wormer- en Jisperveld’ (52°29′58″N, 4°48′57″E). Within each area, as wide a range as possible in land-use intensity was sampled. Only fields that had not yet been mown or grazed were selected.

Vegetation-inhabiting arthropods were sampled with emergence traps (‘photoeclectors’). On each field, three traps were placed that each sampled 1 m2 of vegetation. Traps were emptied on a weekly basis, after which the traps were placed on a new spot in the same field. Sampling started on 11 May 2006 in the Gerkesklooster area and 12 May 2006 in the Ronde Hoep area, and traps were open for a total of 6 weeks. In the Wormer- en Jisperveld area sampling started on 3 May 2007 and lasted 5 weeks. Arthropods were sorted to order level and counted, and dry weight was determined after 72 h at 70 °C. For the most abundant order, the Diptera, three size groups were distinguished when counting individuals (individuals < 3, 3–7 and > 7 mm body length) but not when determining dry weight.

We determined the vertical structure of the vegetation by taking digital photographs at the start of sampling and each time a trap was emptied. Photographs were taken within 2 m of each trap. A white board (60 cm high, 70 cm wide) was placed vertically in the vegetation. Photos were taken 0.5 m above ground level at a distance of 1 m from the board. On the photos, vegetation contrasted darkly against a white background. Using the sidelook software (Zehm et al. 2003) we determined for each trap and sampling date the percentage vertical vegetation cover at 5-cm height classes. We subsequently calculated for each trap and date at what height the vegetation had 50% vertical cover. Pooling the data from the three traps we then calculated per field the vegetation height with 50% cover on 16 May.

Data on typical fertilizer use on the fields were obtained from farmers using questionnaires. Nitrogen (N) input from various types of organic fertilizers was calculated using Van Dijk (2003) to obtain the total amount of N applied per field in kg /ha/yr.


To analyse how arthropods were related to the land-use intensity parameters we used generalized linear models (GLMs). Because the explanatory variables (fertilizer input and vegetation height) acted at the field level, data from different traps and dates were averaged per field. As abundance data generally do not follow a normal distribution, they were analysed using log-linear models employing the Poisson distribution and using a log-link function (McCullagh & Nelder 1989). Overdispersion was accounted for by inflating the variance of the Poisson distribution with a constant factor and assessing fit of the model using test statistics that assume F-distributions (Payne et al. 2002). Arthropod dry weight and mean body weight data were analysed using GLMs employing a normal error distribution and using a log-link function followed by F-tests. To correct for differences in sampling effort, climatic and environmental conditions between years and areas, the factor ‘area’ (Gerkesklooster, Ronde Hoep, Wormer- en Jisperveld) was included in all statistical models. Models therefore contained the correcting factor variable ‘Area’ as well as the explanatory variables ‘Vegetation height’, ‘Fertilizer input’ and their interaction. Because the explanatory variables were correlated (= 0.412, P = 0.029) the effect of one of the variables in statistical models may depend upon the presence of the other. We therefore examined the effects of the variables in the two alternative models. Statistical results are reported for the model consisting of Area + Vegetation height + Fertilizer input + Height*Fertilizer, but when significance of effects depended on the order in which the variables enter the model, this is indicated. All analyses were undertaken in genstat 8.11 (Payne et al. 2002).


Changes in timing of agricultural activities and breeding Godwits

Between 1982 and 2005, agricultural activities in spring moved forward by approximately 15 days (Fig. 1a; F1,31 = 8.14, P = 0.008). The temperature sum at which the median mowing date occurred did not change significantly between 1982 and 2005 (Fig. 1b; F1,31 = 1.65, = 0.209).

Figure 1.

 The relationship between (a) Julian date and (b) temperature sum and median mowing date (solid lines) as well as hatching date of Black-tailed Godwit chicks (hatched lines) in a range of Godwit breeding sites on agricultural grasslands between 1982 and 2005. Bold lines indicate relationships with slopes significantly different from 0, thin lines relationships with slopes not significantly different from 0. Data are from Kruk et al. (1996), Schekkerman (1997), Schekkerman et al. (1998), Teunissen (2000) and Schekkerman et al. (2005).

Over the same time period and in the same study areas, median hatching dates of Black-tailed Godwit chicks did not change (Fig. 1a; F1,22 = 0.040, P = 0.841). As a result, the median Godwit chick in 1982 hatched approximately 11 days before the median mowing date, whereas in 2005 it hatched 2 days after the median mowing date. The temperature sum at the median hatching date increased significantly by approximately 210 °C (Fig. 1b; F1,22 = 8.84, P = 0.007) and showed a temporal trend similar to that of the temperature sums on 1 May and 1 June.

The recent stability of Godwit hatching dates is corroborated by the national ringing data of chicks, which indicate no change or a slight delay since the 1970s (Fig. 2a). However, in the three decades before 1975, chick ringing dates advanced by about 2 weeks. Prior to 1950 data are scarce but suggest a constant hatching date or at most a slight advancement (Fig. 2a). The development of Tsums in spring did not change during the greater part of the past century but has shown a marked acceleration since the 1970s (Fig. 2b). Hence, Godwits bred at progressively lower Tsums between the 1940s and the 1970s, approximately keeping up with the advancement of reconstructed first mowing dates in this period (Beintema et al. 1985). Since about 1980, however, breeding phenology has not kept up with the warming of spring weather whereas mowing dates have, and hatching occurs at progressively higher Tsums (Fig. 2b).

Figure 2.

 Changes in the breeding phenology of Black-tailed Godwit in the past century based on ringing dates of chicks throughout the Netherlands (a), and in relation to the development of spring climate (b). In (a), black symbols show the median ringing dates of chicks (squares are means over 6–12-year periods as given by Beintema et al. 1985, n = 18–582 chicks per period); dots are annual medians (n = 125–1632 chicks/yr). The bold black line is a smoothing spline (df = 5) through the hatching dates, estimated assuming that the average age at ringing was 13 days until 2002 and 4 days in 2003–2006 (see Methods); median first egg dates will have fallen c. 27 days earlier (thin black line). Also shown are approximate peak arrival dates (thick grey line) based on Mulder (1972) for the 1930s and 1960s, and on first spring resightings of Godwits colour-ringed in the western Netherlands in 2004–2007 (H. Schekkerman unpubl. data). The broken grey line shows dates of first grass cut reconstructed by Beintema et al. (1985) for 1910–1970, linked to results of this study in 1982–2005. In (b), the development of winter and spring climate is shown as temperature sums at the end of each month (grey smoothing splines, df = 5) Temperature sums on the median hatching date of Godwit chicks (from a) are shown by dark grey symbols and a black smoothing spline (df = 5).

The relationship between land-use intensity and quality of chick foraging habitat

In total, 268 188 arthropods were caught. Diptera was by far the most abundant order with slightly over 50% of all individuals and 43% of the total dry weight (Table 1). Eighty per cent of the Diptera were smaller than 3 mm and only 3% were larger than 7 mm. Araneae and Hymenoptera made up a significant proportion of the number of arthropods and, to a lesser extent, total dry weight. Coleoptera and Hemiptera were present in small numbers but, due to their large mean body weight, Coleoptera contributed considerably to the total dry weight (Table 1).

Table 1.   Mean number, dry weight and body weight of different arthropod orders emerging from wet grassland vegetation in the chick-rearing period of Black-tailed Godwits in the Netherlands.
 AbundanceDry weightBody weight
 < 3 mm243.341.1   
 3–7 mm50.78.6   
 > 7 mm10.11.7   

Vegetation height was generally positively related to the number and dry weight of arthropods (Table 2, Fig. 3a). Relationships were highly significant for Diptera, Araneae and Hemiptera. Coleoptera and Hymenoptera were not related to vegetation height, nor was the summed dry weight of all arthropod orders. Vegetation height was negatively related to individual body weight of all arthropod groups, but this relationship was significant for Araneae and Hemiptera only.

Table 2.   The effects of vegetation height (at 16 May) and fertilizer input on the numbers, dry weight and mean body weight of different orders of arthropods in wet grasslands. For clarity, the effects of the correcting factor ‘Area’ that was included in the analyses are not shown.
 Vegetation heightFertilizer inputHeight*Fertilizer
  1. †Not significant when included as first explanatory variable in statistical model.

  2. ‡Not significant when included as second explanatory variable.

  3. §Significant when included as first explanatory variable.

 Diptera+15.20< 0.001+0.040.8522.140.158
  < 3 mm+13.530.001+0.080.7753.100.092
  3–7 mm+10.220.0042.180.154+0.030.860
  > 7 mm+1.600.2191.590.2210.170.681
 Araneae+19.57< 0.001+23.97< 0.00112.360.002
 Hymenoptera+0.000.984+19.64< 0.0016.480.018
 Hemiptera+29.48< 0.001+0.090.7663.840.063
 All orders+6.470.018+9.370.006†5.340.031
Dry weight
 Araneae+15.23< 0.001‡+16.22< 0.0012.850.105
 Hemiptera+17.51< 0.001+0.110.7441.070.311
 All orders+0.810.378+3.880.0620.910.350
Mean body weight
 Araneae22.49< 0.0013.180.088§+7.850.010
 All orders0.760.3940.010.9391.320.263
Figure 3.

 An illustration of the relationship between vegetation height and (a) the number of arthropods per area and (b) the number of arthropods per volume of vegetation. This latter analysis is based on the conservative assumption that the density of vegetation with < 50% cover remains constant with increasing vegetation height. The line in (b) shows a significant exponential relationship (F2,450 = 130.60, P < 0.001, R2 = 36.4) that gave a better fit than a linear relationship (F1,451 = 169.32, < 0.001, R2 = 27.1).

Fertilizer input was significantly positively related to Araneae abundance and dry weight and Hymenoptera abundance. Furthermore fertilizer input was positively related to total arthropod numbers and dry weight, although the relationship with numbers was only apparent when the relationship with vegetation height had been accounted for (Table 2). The relationships between fertilizer input and mean body weight were all negative but none was significant. The negative effect of fertilizer on Araneae mean body weight was largely explained by the confounding effects of vegetation height.

Interactions between the effects of vegetation height and fertilizer input were mostly not (or marginally) significant. Significant interactions all indicated that the effect of one variable declined with increasing levels of the other variable (Table 2; opposing signs of effects of main variables and interactions).


Changes in timing of agricultural activities and breeding Godwits

During the last 25 years, Black-tailed Godwits hatched their chicks at approximately the same date, whereas first mowing and grazing dates advanced by approximately 15 days. Perhaps surprisingly, our results suggest that the advancement of farming activities could largely be explained by climate change rather than by agricultural intensification. In 2005, farmers cut or grazed their grasslands at the same temperature sum, and therefore the same developmental stage of the vegetation as in 1982, the first year of this study. This temperature sum was, however, reached significantly earlier in the year. A possible explanation for the lack of any evidence of an advance of agricultural activities not caused by climate change may lie in the fact that agriculture in the Netherlands was already very intensive in the 1980s. Fertilizer inputs have actually decreased by 20% since then (Kleijn et al. 2009a), although other aspects of intensification like drainage, reseeding of grassland, and technical improvement of mowing and harvesting gear have continued.

In contrast to the advancement in recent decades, the advancement of mowing dates prior to 1980 probably does reflect agricultural intensification, as it occurred during a period of stable spring climate (Fig. 2). Lowering of water tables and increased fertilizer application stimulated grass growth independently of weather and the switch from hay to silage making led to swards being cut at an earlier stage. These developments probably also enabled (through an increase in soil macrofauna as food for adults or earlier availability of nest cover) and/or forced (through selective losses of later-hatching eggs and chicks) Godwits to considerably advance their breeding period (Beintema et al. 1985).

It is unclear why Black-tailed Godwits did not continue to advance their laying date after the 1970s. Selection pressure favouring earlier breeding still exists given that late-laying Godwit pairs produce considerably fewer fledglings than early pairs (Roodbergen & Klok 2008). Furthermore, the date on which the first Lapwing Vanellus vanellus eggs of the season were found by egg collectors advanced by about 10 days in the second half of the 20th century (Both et al. 2005). This was linked to higher spring temperatures and higher winter rainfall (Both et al. 2005). It is unknown what cues Godwits use to start egg-laying and incubation. The spring increase in temperature and start of vegetation growth have continued to advance in recent decades and are therefore unlikely to have formed constraints, which suggests an inability of adults to advance their arrival from the wintering grounds (Both et al. 2005) or nutritive constraints in the pre-laying period (Högstedt 1974, Nager 2006) as possible causes. The scant data available indicate that arrival dates of Godwits in the Netherlands have advanced and then stabilized roughly in parallel with laying dates, and that the length of the pre-laying interval has not changed (Fig. 2a). Understanding the lack of response of Black-tailed Godwit phenology to environmental change therefore may require more information on their spring migration and on foraging and energetics during the pre-laying period.

The relationship between land-use intensity and quality of chick foraging habitat

This study used vegetation height at median hatching date and N input as proxies for land-use intensity. It is unclear how well these proxies capture the wide range of aspects that are affected by agricultural intensification. However, vegetation height before the first cut has the advantage over other indicators, in that it is the product of all the activities taken by the farmer to optimize grassland productivity and it is relevant from the perspective of a foraging Godwit chick. Nitrogen input is less directly related to vegetation productivity or chick foraging habitat, but is used in many studies as an indicator of land-use intensity (e.g. Herzog et al. 2006, Kleijn et al. 2009b), which allows us to compare our findings with these studies.

The relationship between arthropods and the land-use intensity indicators was dominated by the response of the Diptera, which made up between 43 and 52% of all arthropods. This is similar to the range found by Schekkerman and Beintema (2007), who used the same methods in Dutch wet grasslands in the 1990s. Their abundance may explain why Dipterans are the most frequently taken prey items by Godwit chicks (found in 94% of the faecal samples; Beintema et al. 1991) rather than there being a genuine preference for this insect order. Ground-dwelling Coleoptera are not considered to be important food items, as Godwit chicks generally take food items from the vegetation rather than from the ground, and the Coleopterans in our samples were mainly ground-dwelling taxa such as Carabidae and Staphylinidae. Combining the arthropod abundance data (Table 1) with chick foraging preferences therefore suggests that in Dutch grasslands, Diptera, Araneae and Hymenoptera are the most important groups for Black-tailed Godwit chicks in terms of food quantity.

Schekkerman and Beintema (2007) found a pronounced short-term decline in arthropod abundance when fields were cut, which is relevant to Godwit chicks, as the first cut increasingly overlaps with their hatching period. However, Godwit broods strongly prefer to stay in uncut fields and travel considerable distances to reach them (Schekkerman & Beintema 2007), and hence variation in food availability in uncut fields as measured in this study is highly relevant.

In such uncut grasslands we found that, when significant, the relationships between the abundance of different arthropod groups and our two indicator variables of land-use intensity were all positive. The relationships of vegetation height and fertilizer input with total arthropod abundance were, however, only marginally significant and the relationships with total arthropod dry weight were not statistically significant, suggesting that the impact of agricultural intensification on the total prey abundance of Godwit chicks was positive but not very pronounced. Effects of vegetation height and fertilizer input on mean body mass were generally negative, although only convincingly so for Araneae. This is in line with the findings of Siepel (1990) and Blake et al. (1994) who found declining arthropod body size with increasing management intensities in grasslands. More intensively managed grasslands therefore contain more, but slightly smaller, prey items.

The body size of prey items is important for Godwit chicks because it determines how efficiently they can forage. Arthropods with a body dry weight below 1 mg cannot be used profitably (Schekkerman & Boele 2009), particularly by older chicks, and Godwit chicks indeed seem to prefer large arthropods (Beintema et al. 1991). In the examined grasslands, only Coleoptera and some Lepidoptera (which were not analysed separately but were included in the analyses of the total of all orders) consisted of animals in this size class. The Coleoptera consisted mostly of Carabidae and Staphylinideae, which are rarely eaten by Godwit chicks (Beintema et al. 1991), whereas mean body weights of the most abundant orders (Diptera, Araneae and Hymenoptera) were all well below 1 mg. Illustrative of this is that about 80% of the Diptera consisted of very small individuals (< 3 mm). If this size distribution is representative for the other arthropod groups, it suggests that total arthropod abundance may be a poor predictor of habitat quality, as only a small proportion of the total arthropod abundance can be profitably used by Godwit chicks. Interestingly, the arthropod groups with large individuals (Coleoptera, Diptera > 7 mm body length) were not significantly related to land-use intensity indicators.

The effects of land-use intensity on the quality of foraging habitat of Black-tailed Godwit chicks is therefore probably unrelated to food abundance. This suggests that, similar to findings for Meadow Pipits Anthus pratensis in UK grasslands (Vandenberghe et al. 2009), accessibility and harvestability of food may be the key factors determining the quality of the foraging habitat of Godwit chicks. Foraging efficiency of many grassland birds decreases as vegetation height increases because of a reduction in forager mobility and prey detectability (Butler & Gillings 2004). In our study, a large proportion of the fields were characterized by vegetation with 50% cover at 30 cm or higher during the Godwit chick period (Fig. 3a). These fields may be very difficult to access by Godwit chicks, particularly when they are young, unless they have an open horizontal structure, which is usually the case only at low levels of land-use intensity. Furthermore, because vegetation height increases more rapidly than arthropod abundance, tall swards host lower densities of arthropods (Fig. 3b). From the perspective of Godwit chick foraging efficiency, fields with low or sparse vegetation will therefore be most suitable. In such fields, the energetic cost associated with foraging is lowest and prey items are easiest to detect because they are concentrated in a small volume of vegetation.

Conservation implications

Over the last few decades, farming activities in the Netherlands advanced but Black-tailed Godwit hatching dates did not. Twenty-five years ago most Godwit chicks hatched well before the onset of mowing and grazing. Currently, the hatching peak occurs after most mowing activity has taken place. This has a number of important implications. First, it means that an increasing proportion of clutches and vulnerable 1–2-day-old chicks are exposed to mowing activities or grazing cattle. This may explain why losses of Godwit clutches caused by agricultural activities were 6.7 times higher in the late 1990s than in the late 1980s (Teunissen et al. 2005). Secondly, hatched chicks currently find themselves in landscapes that are largely devoid of their preferred foraging habitat: tall uncut grasslands. Schekkerman et al. (2009) showed that the predation hazard for Godwit chicks is higher in recently cut or grazed fields than in uncut grasslands. Thirdly, the vegetation on the remaining uncut fields is taller, denser and therefore less suitable for foraging at the time that chicks hatch. This study shows that arthropod abundance is little affected by the vegetation structure and suggests that before the first cut of the season, quality of the foraging habitat is primarily determined by the accessibility of the vegetation to chicks. The decline in availability and quality of preferred foraging habitat may explain why Godwit chicks are currently in poorer condition than in the 1980s and why reproductive success and especially chick survival have declined in this period (Schekkerman et al. 2008). Thus, climate change and the associated advance of the start of agricultural activities result in a cascade of changes to the chick-rearing habitat of Black-tailed Godwits that all adversely affect chick survival.


This study suggests that in recent decades climate change is an important driver of the advance in agricultural activities which, in turn, has caused serious deterioration in the survival rate of Black-tailed Godwit chicks. To our knowledge, this is one of the first studies describing in some detail effects of climate change that arise indirectly through changes in human land use, rather than through direct effects of climate parameters on reproduction or survival of organisms. Agri-environment schemes delaying the first agricultural activities are currently in place in most key breeding sites of Black-tailed Godwits on agricultural land. So far, they generally have failed to produce the reproductive success required for a stable population and therefore have failed to stop the decline of Godwit populations. Our results suggest that delaying the first mowing or grazing date is currently less effective as a conservation tool than 25 years ago because higher winter and spring temperatures induce a more rapid vegetation growth, resulting in denser, less accessible and therefore lower quality vegetation in the chick-rearing period. For agri-environment schemes to really make a difference, they should be implemented in a larger percentage of the breeding area than the 20–30% that is now generally in place (Schekkerman et al. 2008). However, our findings indicate that even this may not have the desired effects unless the quality of the uncut vegetation is improved by measures that create open, accessible swards. This can most effectively be done by raising groundwater to levels just below the field surface throughout the chick period. The resulting oxygen deprivation of plant roots will retard vegetation growth (Jackson 1985) independently of temperature or fertilization level, thus creating open swards even in productive grasslands and warm springs.


This study was funded by the ‘Kenniskring weidevogellandschap’ and by the ministry of LNV (project BO-02-002/003). We thank all farmers, reserve managers and the ‘Vereniging Natuurmonumenten’ for their help and willingness to let us carry out this study on their land.