Study Area

We recorded foraging success at two study areas in 1987 and at five areas in 2014, scattered throughout the region of intensive rice cultivation in northwestern Italy (Table 1; Fig. 1). One of the areas was the same for both study periods, but the second 1987 area could not be resampled because of changes in land use in 2014. Therefore, during 2014 we were compelled to adopt new study areas within the relatively uniform region of rice cultivation, where the paddies cover large portions of the landscape. Each study area was within a flight distance of 5–10 km from one to three heronries.

In 2013–2014, we replicated the field techniques used from 1985 to 1988 to investigate the foraging ecology of the three study species, so as to allow a comparison between the two study periods. The original data for 1985 (Fasola 1986) and for 1987 and 1988 (M. Fasola, unpubl. data) were clumped and are hereafter referred to as “1987″ data. The 1987 data were re-analyzed together with the new data for 2013–2014, which also were clumped and are hereafter referred to as “2014″ data. The aim of assessing the long-term changes in the average foraging success justifies clumping the results for subsequent years into the two periods “1987″ and “2014″ spaced across three decades. The possible inter-annual differences were taken into account in the models.

Data Collection

Each study area was surveyed throughout the daylight hours, from 06:00 to 19:00 hr, once per week from 3 May to 22 June in 1985, 1987 and 1988, and from 7 May to 30 June in 2013 and 2014. This was when the proportion of nests with chicks was at its peak in the nearby heronries. We approached each adult bird that was actively foraging and in breeding plumage. Focal birds did not increase distance from the observer or appear to change behavior. A focal observation was performed on each bird by a single observer from a car at a distance from 30 to 80 m (the maximum distance of the birds in the paddies from our path). All observed focal birds were foraging in flooded or semi-flooded rice paddies that had been temporarily drained for cultural operations and maintained a partial water cover. In most cases, focal observations were conducted for 10 min (n = 532, 90.5% of observed birds) using spotting scopes (20–60x); in a few preliminary trials, sampling lasted more than 10 min (n = 56, 9.5% of observed birds). During each observation we recorded number, type, and size of captured prey. When visual identification of small prey type was dubious, their identity was confirmed at the end of each observation by net sampling at the spot where the focal bird had been feeding. Prey size was estimated in the field as a proportion of bird bill standard length, and was later calculated from the mean bill length values: 120.3 mm for Grey Heron (Cramp and Simmons 1977), and 72.4 mm for Blackcrowned Night-Heron and 90.4 mm for Little Egret (measured on 12 and 16 individuals of both sexes, respectively, captured within our study area from 1989 to 2014). One of the observers collected records in each study year. All observers attended training sessions at the beginning of each year to ensure consistency in prey identification and size estimation.

Table 1.

Study areas within the region of intensive rice cultivation in northwestern Italy.

For each focal bird observation, we calculated the intake of each prey type as biomass (g, dry mass/min). Biomass reflected the nutritional value more accurately than the number of items because the different prey types ranged in size over several orders of magnitude. We assigned a mass value to each captured prey item by using one of the three following estimates (details in Fasola and Cardarelli 2015). For small to medium size prey that had been identified to species or to the categories of Table 3, we assigned a mean mass based on a large sample of individuals of the same taxon; these prey samples were either obtained in the paddies after the focal observations or obtained whole from regurgitations by nestlings of the same heron and egret species within our study area and were then dried in an oven at 60 °C until a constant weight was obtained. For fish and other large prey, whose length had been recorded during the focal observations, we assigned the mass calculated from length-mass relationships, which were obtained from publications on the local fauna or calculated from sample specimens collected in the rice paddies during our study. The remaining unidentified small items were assigned a combined total mean dry mass based on all of the small items identified for the same heron species.

Figure 1.

Study areas (gray circles, numbers as in Table 1) sampled for foraging success of Grey Herons, Blackcrowned Night-Herons and Little Egrets in northwestern Italy. Dotted line: region of intensive rice cultivation; open circles: heronries; black lines: major rivers.

Data Analysis

We divided the foraging herons and egrets into two categories based on capture success: “successful” individuals that captured > 0 prey and “non successful” individuals that did not capture any prey during the 10-min focal observation. Differences in the percentage of successful and non-successful individuals between 1987 and 2014 were tested by Chi-square with Yates correction, or by Fisher's exact test when the number of observed animals was small (Fisher 1922; Yates 1934; Crawley 2007).

The difference in the number of captured prey items between the two study periods was compared for each prey category separately, using Generalized Linear Mixed Models (GLMMs) with a Poisson distribution and a log link function. The Poisson distribution is typically used for count data, and the log link function ensures positive fitted values. When necessary, we corrected for overdispersion of the data using a negative binomial distribution, that is again used for integers, non-negative data, when the variance is larger than the mean (Zuur et al. 2009). The study period was fitted as a fixed categorical factor (two levels: 1987 vs. 2014), while the site and the year of sampling were fitted as random factors to incorporate the dependency among observations carried out in the same site and in subsequent years. Log of the duration of the focal observations in minutes was used as an offset to account for the different length of each observation (Zuur et al. 2013). We compared the parsimony of the models with and without the year of sampling using the Akaike information criterion (AIC), and we removed the factor when the model was not improved (Burnham and Anderson 2002). For the infrequent prey (number of records < 15), and particularly for those that were not recorded in one of the two study periods, it was not possible to run the models due to the small number of records. The difference in the total biomass intake per minute from rice paddies between 1987 and 2014 was tested by Linear Mixed Effects (LME) models with the same structure of GLMMs described earlier. The biomass intake data were normalized by square-root transformation. All statistical analyses were performed using the statistical software R (R Development Core Team 2016), and the supplementary packages “car” (Fox and Weisberg 2011), “nlme” (Pinheiro et al. 2016) and “lme4″ (Bates et al. 2015).

Energy Budget Estimates

Energy consumption and food demands of free-living birds can be predicted from body mass (Carey 1996; Blem 1999). We calculated the Basal Metabolic Rate (BMR) for our three study species using the equation for non-passerines: BMR (in kcal/bird/day) = 0.522 * M (body mass in g) ^ 0.735 (Kendeigh et al. 1977). For M, we used the values of 1,433 g for Grey Heron (Cramp and Simmons 1977), 580 g for Black-crowned Night-Heron and 505 g for Little Egret. The Daily Energy Expenditure (DEE) is the sum of BMR plus the energy needed for physiological processes, muscular work, and tissue production; for DEE, we used the estimate of 3.6 * BMR, a value recorded in various species of breeding birds in the wild and over extended periods (King 1974; Blem 1999; Hodum and Weathers 2003). In addition to its own DEE, a breeding adult also needs to meet further demands from its nestlings. Therefore, we estimated the peak demands of a nestling from the equation for Peak Daily Metabolized Energy (PDME), which scales with body mass at fledging: PDME (in kj) = 5.01 M (in g) ^ 0.782 (Weathers 1992), using observed M values for Black-crowned Night-Heron nestlings at fledging in our study area (Prigioni et al. 1985). We obtained the number of nestlings from data on breeding success recorded at several heronries within our study area (Fasola 1998; E. Cardarelli, unpubl. data); the values for each study period, calculated as the mean of a sample of 19–53 nests for each species and year, were: 2.3 chicks/nest (1987 and 2014) for Grey Herons; 2.5 chicks/nest (1987 and 2014) for Black-crowned Night-Herons; and 3.5 (1987) and 3.3 chicks/nest (2014) for Little Egrets. The energetic value of each prey category was measured by bomb calorimetry of samples collected within our study areas (Table 2). A metabolized energy coefficient equal to 82.5% of the energetic value was adopted so as to account for digestive and assimilation efficiency (Causey 1999).

Finally, we calculated the amount of prey that a breeding adult Grey Heron, Black-crowned Night-Heron and Little Egret should collect to meet the peak daily energy expenditure for itself and its brood during nestling growth, using the above estimates of energy need, the mean number of chicks/nest in 1987 and in 2014, and the mean energetic content of prey taking into account the different proportion of prey types captured in 1987 and in 2014, as well as assuming a balanced biparental care (Fasola 1984; Hafner et al. 2002).

Table 2.

Energy content of the main prey categories. The values are the mean of three samples, each made by one item for larger prey, or by a composite of several individuals for worms, arthropods and tadpoles.

Prey Type

The prey types in 2014 differed considerably from those in 1987 (Table 3). The primary difference was that new prey species appeared in 2014. The mole cricket (Gryllotalpa sp.) and two allochthonous species, the Louisiana crayfish (Procambarus clarkii) and the oriental weatherfish (Misgurnus anguillicaudatus), invaded our study area after 1990. In addition, worms and small crustaceans belonging to the genus Triops were more frequent in 2014, while frogs and tadpoles were less frequent. The larvae of large aquatic insects were not recorded in 2014, although smaller larvae could have been present among the non-identified small prey (Table 3).

Food Intake

The proportion of successful individuals that had captured at least one prey during focal observation decreased from 1987 to 2014 for Grey Herons (82% in 1987 vs. 55% in 2014) and for Black-crowned Night-Herons (71% in 1987 vs. 43% in 2014). These values differed significantly for both species (χ2 = 6.33, df = 1, P = 0.007 and χ2 = 13.45, df = 1, P < 0.001, respectively). However, for Little Egrets the proportion of successful individuals was the same in both periods, 92% (χ2 = 0.001, df = 1, P = 0.98).

Table 3.

Prey captured in rice paddies by the three study species during the two study periods. Values are percentage of occurrence of each taxon. Insects and amphibians are metamorphosed, if not otherwise specified.

The capture rate by the three studied species (Table 4; Fig. 2) differed significantly between 1987 and 2014 for the main staple prey, even if at the edge of significance in some cases. Frogs were captured at slower rates in 2014 by the three species, particularly by Grey Herons and Little Egrets. The same tendency is also evident for tadpoles. On the contrary, the capture rate of fish was higher in 2014, especially for Little Egrets. The total biomass intake was much lower in 2014 than in the previous period for both Grey Herons and for Black-crowned Night-Herons, while it was similar for Little Egrets (Table 5).

Energy Budget

The total amount of prey that a single individual needed to meet the collective Peak Daily Energy Expenditure for reproduction was estimated for Grey Herons at 207 g (1987) and 216 g (2014) dry mass/day, for Black-crowned Night-Herons at 104 g (1987 and 2014) dry mass/day, and for Little Egrets at 115 g (1987) and 112 g (2014) dry mass/day. From these food requirements and from the observed values of prey intake (Table 5), we calculated the time needed by a bird foraging only in the flooded paddies to satisfy its peak daily requirement for itself plus its brood. One breeding Grey Heron would have satisfied its requirement by foraging for 12 hr per day in 1987, while the diminished prey intake was largely insufficient to meet requirements in 2014 when 51 hr per day would have been necessary. The intake by Black-crowned Night-Herons was already insufficient to meet the daily requirements in 1987 when one bird would have to forage for a non-sustainable 22 hr per day, and it was even more insufficient in 2014 when 35 hr would have been required. Only the relatively higher prey intake by Little Egrets could have allowed individuals to meet the daily requirements of a breeder by foraging for 16 hr per day in 1987 and for 14 hr per day in 2014.

Table 4.

Estimates (± SE) of intercept and covariate coefficients for the GLMMs comparing prey captured by the three studied species in 1987 and 2014. A separate model was run for each prey type (df always equal to 1); intercept estimate represents prey captures in 1987 while covariate estimate represents prey captures in 2014.

Figure 2.

Capture rate (mean number of captured items/min ± 95% CI) in the two study periods (white bars for 1987, gray bars for 2014) for the prey types whose capture rate differed at P < 0.1.