|The importance of rice ﬁelds for glossy ibis (Plegadis falcinellus): Management recommendations derived from an individual-based model
Gregorio M. Toral a,⇑, Richard A. Stillman b, Simone Santoro a, Jordi Figuerola a
a Department of Wetland Ecology, Doñana Biological Station, CSIC, Avda. Américo Vespucio s/n 41092, Seville, Spain
b School of Applied Sciences, Bournemouth University, Talbot Campus, Poole, Dorset BH12 5BB, UK
Keywords: Wetland Agriculture Crayﬁsh Waterbird MORPH
a b s t r a c t
Artiﬁcial wetlands provide alternative habitats for waterbirds. The Doñana rice ﬁelds (SW Spain) are extensively used as a foraging site by the glossy ibis (Plegadis falcinellus). The aim of this study was to develop an individual-based model to predict the possible effects of glossy ibis’ population growth, reductions in the rice cultivated area, and changes on the phenology of the management processes of the paddies on the mortality rate of the glossy ibis population. We test the hypothesis that the glossy ibis breeding population of Doñana can obtain its energy requirements during the non-breeding season by feeding on rice ﬁelds alone. Our results show that the glossy ibis population growth is not currently lim- ited by rice ﬁeld availability. However, a reduction of 80% would cause mortality rate increases above the observed mortality (5.9% per year), with values around 10% per year for populations between 50,000 and
100,000 birds. A reduction of 90% of the rice ﬁeld area would cause mortality rate increase above the observed value for populations greater than 20,000 birds, reaching 60% per year with a population of
100,000 birds. Cultivated area in Doñana suffers temporary reduction on its area during drought periods. Taking into account the fact that the glossy ibis population for 2011 may exceed 22,900 birds, large scale changes in the area of rice ﬁelds due to habitat transformations and/or drought periods may have impor- tant effects on the viability of the glossy ibis population in Doñana.
In the current scenario of loss of natural marshes, the surface area of artiﬁcial wetlands has increased in many areas, providing alternative habitats for waterbirds (Elphick and Oring, 1998; Elphick, 2000; Tourenq et al., 2001; Ma et al., 2004). Rice ﬁelds have been proven to be of great value for waterbirds in different regions of the world (Fasola and Ruiz, 1996; Elphick, 2000; Maeda,
2001; Czech and Parsons, 2002; Sánchez-Guzmán et al., 2007; Toral and Figuerola, 2010) and are thus considered the world’s most important agricultural crop for waterbirds (King et al.,
2010). Furthermore, in areas suffering drought periods (like the Mediterranean basin) this crop could play an important role as a refuge for waterbirds when natural marshes dry (Tourenq et al.,
2001; Rendón et al., 2008). However, since the scarcity of water in those areas could also threaten the viability of rice ﬁelds, it is important to understand how future reductions in the area of rice ﬁelds could affect waterbird populations.
In the northern hemisphere the glossy ibis (Plegadis falcinellus) has a near cosmopolitan distribution, including areas in Africa, America, Asia, Australia and Europe (Figuerola et al., 2004). In the
⇑ Corresponding author. Tel.: +34 654442904; fax: +34 954621125.
E-mail addresses: firstname.lastname@example.org (G.M. Toral), email@example.com
Iberian Peninsula the glossy ibis became extinct as a breeding spe- cies at the beginning of the 20th century. Three failed recoloniza- tion attempts of the area by glossy ibis occurred in Doñana marshes during 1930–1935, 1940–1942 and 1958 (Figuerola et al., 2004). New breeding colonies were established in the marshes of the Doñana National Park (SW Spain, Fig. 1) and Ebro Delta (Catalonia, NE Spain) in 1996 (Santoro et al., 2010). The col- ony of Doñana has rapidly increased (Fig. 2) to become the largest colony in Western Europe (Figuerola et al., 2004) and held more than 3500 pairs in 2007 (Máñez and Rendón-Martos, 2009), and about 5300 pairs in 2010 (Máñez pers. com.). However, the con- centration of this important breeding colony in a few locations in the Doñana marshes increases the risk of extinction due to any exceptional event that affects Doñana (for example no breeding occurred in 1999 and 2005 due to sever droughts). At present, the glossy ibis is considered IUCN Vulnerable in Spain (Figuerola et al., 2004) and is a species of conservation concern in Europe, where it is in decline (Delany and Scott, 2006).
Rice is an important food for many avian species (Lourenço and Piersma, 2008; Stafford et al., 2010). Glossy ibis consistently exploit rice agriculture throughout their range (del Hoyo et al.,
1992; Hancock et al., 1992; Taylor and Schultz, 2010). There are a few detailed studies of glossy ibis diet carried outside Europe (Acosta et al., 1996; Davis and Kricher, 2000). The only study in Europe, performed in Doñana, indicates that during the breeding
Fig. 1. Location of the study area showing rice ﬁelds (black) and the boundaries of the Doñana National Park (solid line) and Doñana Natural Park (dashed line). The symbol (*)
indicate the location of the main breeding colony of glossy ibis.
season the glossy ibis diet is dominated by aquatic Coleoptera and Odonata (Macias et al., 2004). However, during the non- breeding season in Doñana, glossy ibises feed on the waste rice grains present in the paddy ﬁelds after the harvest (pers. obs.). This shift in the diet between the breeding and non-breeding season has been also observed in the glossy ibises feeding in rice ﬁelds in Cuba, and conﬁrmed by looking at stomach contents (Acosta et al., 1996).
During the non-breeding season (from September to February), most of the glossy ibis leave the marshes of Doñana and go to sur- rounding rice ﬁelds, where they feed on rice grains in post- harvested ploughed rice paddies. While feeding on rice ﬁelds, glossy ibis use a roost in a well conserved area situated inside the rice ﬁeld area, near the river Guadalquivir. Evaluating the qual- ity of Doñana rice ﬁelds as a foraging site for this species and the possible negative effects of future reductions in the area cultivated on the mortality rates of this growing population is essential to in- form the appropriate management decisions.
Individual-based models, comprised of ﬁtness-maximising indi- viduals, are a potential tool for predicting the mortality rates and body conditions of birds under different scenarios (Stillman et al.,
2005a). Individual-based models based on the MORPH software
(Stillman, 2008) have been successfully used to predict the effects
Fig. 2. Increases in the breeding season population of glossy ibis in Doñana, Southwest Spain (based on Santoro et al. (2010) and unpublished data from the Natural Processes Monitoring Team of the Doñana Biological Station (http://www- rbd.ebd.csic.es/Seguimiento/seguimiento.htm).
of changes in the environment in a variety of species of shorebirds and wildfowl and have successfully predicted changes in mortality rates due to loss of habitat (e.g. Stillman and Goss-Custard, 2010; Stillman et al., 2005b; Durell et al., 2006). MORPH models have been shown to be of great value when assessing conservation objectives of waterbirds (Stillman et al., 2010). The key feature of these models is that they are based on the assumption that individ- uals within animal populations always behave in order to maxi- mise their own chances of survival and reproduction, no matter how much the environment changes (Stillman, 2008). The deci- sions made by model animals are based on optimal foraging theory and game theory, which are thought to provide a reliable basis for prediction (Goss-Custard, 1993; Sutherland, 1996; Goss-Custard and Sutherland, 1997). Therefore, animals in these individual- based models are expected to respond to environmental change in the same ways as real animals would (Goss-Custard, 1993; Sutherland, 1996; Goss-Custard and Sutherland, 1997). Using this model, population level parameters are predicted from the fates of all individuals in the population (Stillman et al., 2005a).
In this paper we demonstrate how a MORPH individual-based model can be used to assess the quality of rice ﬁelds as a foraging site during the non-breeding season for the glossy ibis. The MORPH software has not previously been used to assess the quality of an artiﬁcial habitat. Furthermore, to our knowledge, this is the ﬁrst individual-based model developed for the glossy ibis. We test the hypothesis that the current and future glossy ibis population of Doñana can obtain their energy requirements during the non- breeding season feeding on rice grains in the rice ﬁelds surround- ing the Doñana National Park. We ﬁrst test if the current rice ﬁeld area is able to support an increased population of glossy ibis as, at present, it is still growing (Santoro et al., 2010). We then model loss of habitat and changes in the phenology of the management of the paddies to predict the effect of environmental change on mortality rates. We consider the implications of our results for the conservation of the species in Doñana.
2. Materials and methods
2.1. Study site
The rice ﬁelds of Doñana (Fig. 1) are the largest rice ﬁeld area in Spain. They are situated near the marshes of the Doñana National Park, a 55,000 ha wildlife reserve north of the Guadalquivir estuary
in Andalucía (SW Spain). The climate is Mediterranean sub-humid with rainy winters and dry summers. For more details of the area see Serrano et al. (2006), Rendón et al. (2008) and Kloskowski et al. (2009). Of the 180,000 ha of fresh and brackish marshes pres- ent in 1900, 36,000 ha were transformed into rice ﬁelds between
1926 and 1997 (García-Novo and Martín-Cabrera, 2006; Rendón et al., 2008). Additionally, other types of transformation have reduced the area of natural marshes to 30,000 ha at present (Enggass, 1968; García-Novo and Martín-Cabrera, 2006). Between September and January the habitat structure of rice ﬁelds change quickly as paddy ﬁelds are harvested and thus undergo a series of management processes. A few days after a paddy ﬁeld is har- vested, usually at the end of September, it is ploughed whilst still wet to favour the decomposition of the stubble, which is mixed with the soil. Subsequently, the paddy ﬁeld is ﬂooded with a variable amount of water (depending on the farmer and the amount of rain that has fallen) and stays ﬂooded until it dries up, usually in December–January.
2.2. The model
We used MORPH to develop our individual-based model. As
MORPH has been described in detail elsewhere (Durell et al.,
2006; Stillman, 2008), we limit ourselves to a detailed account of the model’s parameterization for glossy ibis. The parameters were derived from previous literature and newly collected data. A sensi- tivity analysis (see Appendix A) was performed to calibrate the model and to understand the impact of variations in each parame- ter on the model outputs. We optimised model complexity and reduced uncertainty by using multiple patterns observed in the real system at suggested by Grimm et al. (2005). We look at the observed patterns that seem to characterise the system and its dynamics and then included the appropriate variables and pro- cesses so that those patterns emerge from the model. We evalu- ated how the different models (with different parameter values) reproduce observed behaviour patterns of glossy ibis and selected the model that produced the most accurate predictions (Grimm et al., 2005) (see Appendix A). MORPH tracks the location, behaviour, and ultimate fate of each individual in a population and incorpo- rates variation in the foraging abilities among individuals. The model follows the behavioural decisions of each individual as it attempts to meet its daily energy requirements. Individuals that exceed their energy requirements add to their body stores until a maximum is reached. Individuals that cannot meet their require- ments deplete their body stores to survive, but die of starvation if they deplete these stores completely.
2.3. Model parameters
An overview of the equations and default settings of the indi- vidual bird parameters in the model is given in Table 1. The model comprised all of the rice ﬁeld area available in Doñana. Based on the cultivated area in 2005 (37,442 ha), the model encompassed
67 patches, one of them representing the roost and the rest repre- senting 66 IPGs (Integral Production Groups), which are groups of paddies that have parallel control and management under the Inte- grated Production System. Patches size ranged from 53 to 1236 ha) (X = 567 ± 183 ha). The distance from the patches to the roost ran- ged from 227 to 26,499 m (X = 9712 ± 4390 m). We assumed that the initial population consisted of 7000 adult birds (breeding population size in 2007). Only adult birds were considered in the model. We believe that this assumption is appropriate, as the age of ﬁrst breeding in this species at Doñana is 1 year (Máñez and Rendón-Martos, 2009). In practice, to reduce the time taken to run simulations, in the model each forager represented 35–500 identical individuals (super-individuals sensu, Scheffer et al.,
1995). Foragers (but not individuals) differed in their foraging abil- ities and behaviour. The initial size of body store (Table 1) was cal- culated using data from the CRC handbook of avian body masses (Dunning, 1993).
The MORPH model source code and user guide are available as an online appendix to a study that has already been published (Stillman, 2008). The parameter ﬁles used in this study are avail- able as online appendices (Appendix B and C).
2.3.1. Time period
The model simulated glossy ibis feeding on rice ﬁelds during the non-breeding season (150 days) from 20th September, when all the paddies are sown, to 18th February, when all the paddies are dry and birds return to the marshes of the Doñana National Park, which usually have been ﬂooded by winter rainfall (although they can be dry during severe drought periods). The model mimicked how paddies are harvested, ploughed while containing some water, and subsequently dry. The days on which each patch was harvested, ploughed and became dry were randomly selected for each simulation from within the corresponding range of dates ob- served in the ﬁeld during the crop seasons of 2005–2006 and
2006–2007. Glossy ibis migrated to the model following a uniform distribution of arrival dates during the period between the day on which the ﬁrst patch was ploughed and 1st November, the month when the highest number of glossy ibis is observed in rice ﬁelds of Doñana (Toral pers. obs.). Birds emigrated from the model follow- ing a uniform distribution of emigration dates during the period between 10 days after the ﬁrst patch became dry and the day on which the last patch became dry. Paddies became accessible to birds once they were ploughed and remained accessible until they became dry. Birds fed on patches during daylight, and went to the roost during the night. The duration of daylight was calculated from the location of the rice ﬁelds (http://www.usno.navy.mil/ USNO/astronomical-applications/data-services/rs-one-year-us) and ranged from 9 h 35 min to 12 h 15 min. In the model, time was di- vided into discrete 1 h time steps, during which environmental conditions and the distribution of birds remained constant.
2.3.2. Distribution of food supply through the winter
To evaluate the density and biomass of food resources, core samples were taken from 24 ploughed paddies at six locations across the rice ﬁeld area (four replicated paddies on each location). Five replicate cores were taken in each paddy. Samples were col- lected using a 69 mm diameter core, penetrating approximately
10 cm into the mud. To account for seasonal changes in food sup- ply, paddies were sampled three times during the course of winter, October–November 2007, December–January 2008, and March
2008, resulting in a total of 360 cores. Samples were ﬁltered and sieved through 0.5 mm mesh. The number of rice grains per core was counted. We then measured the mean dry mass (after drying for 24 h at 60 °C) and ash-free dry mass (after burning off organic matter for 4 h at 450 °C) of rice grains in each each core. The area of paddies was calculated using ArcGIS 9.2 (ESRI, 2002) from which the density of rice grains was calculated. We used Generalised Lin- ear Models to test whether the location of paddies had an effect on the density and ash-free dry mass of rice grains. We used Wilcoxon tests to explore whether rice grain ash-free dry mass or density varied throughout the winter. Statistical analyses were performed using SAS and SPSS.
Rice energy density was assumed to be 16.51 kj/g (Santiago- Quesada et al., 2009).
2.3.3. Feeding behaviour of glossy ibis
Observations of a total of 56 ringed adults of glossy ibis feeding in four rice paddies between 14 December 2007 and 7 February
2008 were made to measure the mean feeding rate of the species.
Summary of individual bird parameters used in the model.
Number of individuals 200 super-individuals each representing 35–500 identical birds = 7000 to 100,000 individuals
Initial size of body store (g) Drawn from a normal distribution, with a mean of 634 g and an SD of 59 g Target size of body store (g) Drawn from a normal distribution, with a mean of 634 g and an SD of 59 g Starvation body mass 418 g
Intake rate (g s—1) 0.15
Efﬁciency of assimilating resource component Assimilation efﬁciency * efﬁciency to convert carbohydrates into fat = 0.90 * 16.51/33.4
Energy expenditure when feeding or roosting 35 (kj h—1)
Energy expenditure when moving (kj m—1) 4.21
Energy density of bird reserves Amount of energy (kJ) contained in a gram of bird fat reserves = 33.4 kJ g_1 (value measured in shorebirds by Kersten and Piersma, 1987).
Rule for knowledge of site Birds always know food supply in every patch
Survival If body store <418 g at the end of time step, the individual dies.
Each individual was observed over 5 min and feeding observations were recorded onto a digital voice recorder. Birds were observed feeding on rice grains, taking one grain at a time. The formula ap- plied in the model for the functional response of the glossy ibis was deﬁned according to Goss-Custard et al. (2006). Maximum intake rate was based on the maximum daily energy assimilation calcu- lated from body mass using standard equations (Kirkwood,
1983). We used the value of assimilation efﬁciency (90%) corre- sponding to black-tailed godwits (Limosa limosa) eating rice grains (Santiago-Quesada et al., 2009) since glossy ibis’ value is currently unknown. The feeding rate of glossy ibis was determined through a combination of observations during this study and a meta-analysis of the functional responses of shorebirds (Goss-Custard et al.,
2006), a group of birds that are structurally similar to glossy ibis and have similar foraging behaviour. The rate at which glossy ibis feed in the model (FR, measured as rice grains s—1) was determined by the abundance of food in a patch and calculated using the fol- lowing equation (Goss-Custard et al., 2006):
the energy spent to ﬂy to the patch. Model birds then moved to the patch that maximised their energy assimilation. We used a va- lue of 6 h as ringed individual glossy ibis were observed to remain feeding within the same paddy up to 5 h 30 min. We did not use a shorter time as a sensitivity analysis (see Appendix A) showed that this resulted in the model birds avoiding patches furthest away from the roost, whereas these patches were used by the real birds.
2.3.5. Mortality rates
Since the establishment of the colony in 1996, a high proportion (25–80%) of the chicks of glossy ibis have been marked every year with pvc rings. Observed mortality rate (5.9% per year) was indi- rectly estimated by the analysis of capture-mark-resight (CMR) data from 5 years (September 2000–August 2005). We consider that this estimate might be slightly higher than the real mortality rate as this population showed a consistent site ﬁdelity (unpub- lished data) during the period of study (for details, see White and Burnham, 1999). In the individual-based model, the only source of mortality considered was starvation. Illegal hunting is
F ¼ N
other possible cause of mortality for this species in the wild. How- ever, we have not included it in the model since we think that it is a
where Fmax = maximum intake rate when rice grains are superabun- dant, N = rice density in the patch (grains m—2) and N50 = grains density (grains m—2) at which feeding rate is 50% of its maximum.
We used the mean feeding rate measured for glossy ibis in the rice ﬁelds, based on the frequency of swallowing movements (0.15 g s—1) as Fmax, assuming that birds were feeding at their max- imum rate (Goss-Custard et al., 2006). We used the value of N50 cor- responding to redshank (Tringa tetanus) feeding on Hediste diversicolor (46.1 items m—2) (Goss-Custard et al., 2006) since this was the prey species in the meta-analysis with the closest ash-free dry mass (0.006 g) to that of rice grains.418>