Content uploaded by KS Gopi Sundar
Author content
All content in this area was uploaded by KS Gopi Sundar on Sep 03, 2021
Content may be subject to copyright.
Global Ecology and Conservation 30 (2021) e01793
Available online 2 September 2021
2351-9894/© 2021 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Of irrigation canals and multifunctional agroforestry: Traditional
agriculture facilitates Woolly-necked Stork breeding in a north
Indian agricultural landscape
Swati Kittur
a
, K.S. Gopi Sundar
a
,
*
a
Nature Conservation Foundation, 1311, “Amritha”, 12th Main, Vijayanagar 1st Stage, Mysuru 570017, Karnataka, India
ARTICLE INFO
Keywords:
Farmland biodiversity
Habitat suitability modelling
Nest tree preference
Ripley’s K-function
Spatial logistic regression
ABSTRACT
Conservation of biodiversity alongside agriculture is now a global priority. Tree-nesting water-
birds have a tenuous relationship with farmlands because their survival requires farmers to retain
trees and wetlands amid croplands. Research on such birds is rare on tropical and sub-tropical
agricultural landscapes where high human densities and intensive farming ostensibly deterio-
rate breeding conditions. We explored breeding ecology and nest site selection by the single-
nesting Woolly-necked Stork (Ciconia episcopus) in Haryana, north India using 298 nests from
166 locations discovered between 2016 and 2020. We determined the relative strengths of as-
sociation of nest locations with natural features (trees, wetlands), human presence (habitation)
and articial water sources (irrigation canals) to understand cues used by breeding storks to
situate nests. Woolly-necked Stork brood size from 42 successful nests was relatively high (3.1 ±
0.9 SD), with nests close to human habitation and wetlands having smaller broods. Storks showed
high nest site delity (44.5% of sites used >1 year), rarely nested on man-made structures
(electricity pylons; 8.4%), and distributed nests in a clumped pattern. Woolly-necked Storks sit-
uated nests ambivalent to natural features but associated strongly with man-made features
(positively with irrigation canals; negatively with human habitation). Contrary to expectations,
most nests were not on the tallest trees but on the medium sized, native Dalbergia sissoo though
storks situated nests on two tall trees (native Ficus religiosa and exotic Eucalyptus sp.) far more
than the trees’ availability. All three tree species were favoured either for traditional agroforestry
or local religious beliefs. Traditional agriculture in Haryana supported a substantial breeding
population of Woolly-necked Storks facilitated by agriculture-related components rather than
existing natural features. This novel scenario contradicts conventional narratives that suggest
multi-season small holder tropical and sub-tropical agriculture degrades breeding conditions for
waterbirds. Our ndings in Haryana reiterate the need to assemble a diverse conservation toolkit
of different locally relevant mechanisms supporting biodiversity amid cultivation.
1. Introduction
Agriculture is a major cause of the ongoing global biodiversity decline (Tscharntke et al., 2012; Dudley and Alexander, 2017;
Stanton et al., 2018). This recognition has prompted research into ways to retain wildlife amid agriculture since most farmlands are
* Corresponding author.
E-mail address: gopi@ncf-india.org (K.S.G. Sundar).
Contents lists available at ScienceDirect
Global Ecology and Conservation
journal homepage: www.elsevier.com/locate/gecco
https://doi.org/10.1016/j.gecco.2021.e01793
Received 23 July 2021; Received in revised form 1 September 2021; Accepted 1 September 2021
Global Ecology and Conservation 30 (2021) e01793
2
unlikely to be converted to natural habitats given the burgeoning human population. Research is biased towards agricultural systems
in temperate regions where large land holdings, monocultures and mechanization dominate agricultural practices (Tscharntke et al.,
2012; Dudley and Alexander, 2017). It is frequently assumed that trends of biodiversity declines in temperate farmlands are accen-
tuated in small-holder tropical and sub-tropical agricultural systems (holdings ≤2 ha; Ricciardi et al., 2018), especially cereal pro-
ducing farmlands where crop yield is maximised by thinning tree cover (Dudley and Alexander, 2017; Fischer et al., 2017). Emerging
evidence, however, has been contradictory, with small holder tropical and sub-tropical farms with traditional agriculture sustaining far
higher non-crop biodiversity relative to large farms, and cereal farmlands supporting signicant levels of biodiversity despite sparse
tree cover and high human densities (Sundar and Subramanya, 2010; Sundar and Kittur, 2013; Kim et al., 2020; Ricciardi et al., 2021).
Detailed information on requirements of different wild taxa living in crowded agricultural systems is, however, still meagre preventing
a nuanced understanding of how biodiversity may be retained within farmed landscapes globally. Nevertheless, there is rapidly
increasing comprehension that a diverse set of well-informed, locally relevant mechanisms can help with maintaining multifunctional
agricultural systems, which calls for much more research in tropical and sub-tropical agricultural landscapes and on the wild species
found within (Perfecto and Vandermeer, 2010; Fischer et al., 2012; Tscharntke et al., 2012; Koju et al., 2019).
One group of animals that share a tenuous relationship with agriculture are tree-nesting waterbirds (here onwards “waterbirds”; e.
g. herons, ibis, storks). Waterbirds’ ability to persist on cultivated landscapes depends on various factors including the impacts of
altered hydrology to accommodate farming at oodplain scales (Brandis et al., 2018; Bino et al., 2020), the crops grown, especially the
number and kinds of crops grown annually (Tourenq et al., 2009; Fasola and Brangi, 2010; Sundar and Kittur, 2012), attitudes of
farmers towards waterbirds and nest trees (Fasola et al., 2010; Koju et al., 2019), and the extent of natural habitats retained amid
agriculture (Carrasco et al., 2014). In North America, waterbird colonies in human-dominated areas are regarded as a nuisance
requiring removal (Parkes et al., 2012). Such negative attitudes, hunting for sport and food, and large-scale mechanised monoculture
farming that necessitates removal of trees and wetlands has largely relegated waterbird nesting to wetland reserves, protected riverine
forests, and inaccessible islands (Higgins et al., 2002; Kushlan, 2012; Parkes et al., 2012). However, in some locations across Europe
and Asia a patchwork of traditionally managed agriculture exists characterised by favourable farmer attitudes towards waterbirds,
cropland agroforestry with trees utilized for furniture, making yokes and commerce, cultural and religious practices that safeguard
waterbirds and trees, and cropping cycles benecial to waterbird breeding, all of which have led to these landscapes supporting
signicant populations of waterbirds (Tourenq et al., 2009; Sundar, 2011; Koju et al., 2019; Sundar et al., 2019). The expansion of
breeding waterbird studies to diverse agricultural landscapes have provided contestations of existing generalised narratives that
assumed breeding waterbirds require relatively undisturbed wetlands for nesting and would avoid areas such as tropical and
sub-tropical agricultural landscapes with high human population densities. Generalised and widespread narratives unfortunately
continue to dominate discussions, even driving status assessments of waterbird species reecting the relatively sparse research on
waterbird requirements on crowded agricultural landscapes (Sundar and Subramanya, 2010; Sundar, 2020). There is an urgent need to
develop stronger understanding of such multifunctional traditionally farmed landscapes, particularly from Africa and Asia, that host
substantial numbers of waterbird species but have relatively few studies to identify key characteristics that facilitate or deter waterbird
breeding (Gula, 2020; Sundar, 2020).
Existing research on breeding requirements of waterbirds on farmlands has focused on their interactions with two major categories
of variables – persisting natural habitat and human presence. Natural habitats include trees on which waterbirds nest and wetlands that
are important for foraging. Depending on the agricultural landscape and the species, waterbirds show considerable variation in their
reliance on natural habitats for breeding. In some locations, waterbirds situated nests in nearby forests entirely avoiding farmlands
(Nachuha and Quinn, 2010). In other landscapes, waterbirds favoured larger tree patches within or beside farmlands for nesting
(Carrasco et al., 2014). In yet other landscapes, waterbirds used nest trees on farmlands with no apparent relationship with tree
densities on the landscape but favoured larger tree species planted for agroforestry or conserved by religious beliefs (Koju et al., 2019).
Similarly, breeding waterbirds had varying relationships with wetlands on different agricultural landscapes. On some landscapes,
waterbirds favoured nesting close to wetlands or in areas with more wetlands (Fasola et al., 2010). In yet other locations waterbirds
used scattered trees on farmlands to build nests but were ambivalent regarding wetlands around nest sites, likely because farmlands
provided adequate forage (Koju et al., 2019; Sundar et al., 2019). Rarely, waterbirds nested on articial structures such as electricity
pylons in agricultural landscapes. Few studies have investigated this behaviour with existing evidence showing waterbirds prefer-
entially situated such nests near large wetlands and farmlands. This suggests that the same cues could at least sometimes be used to
locate nests irrespective of whether trees or man-made structures were used (Moreira et al., 2018). In addition to natural habitats on
farmlands, articial sources of water such as irrigation canals have long been recognized as important novel habitat for foraging and
breeding of many bird species around the world (Fasola and Ruiz, 1996; Sundar, 2004, 2006; Li et al., 2013; Sueltenfuss et al., 2013;
L´
opez-Pomares et al., 2015; Lin et al., 2020). However, when landscapes retain natural wetlands but also have irrigation canals, it is
not known if waterbirds use both sources of water as cues to situate nests or whether they prefer one over the other.
Choice of nest sites by waterbirds relative to remnant habitat on agricultural landscapes are complicated by human presence.
Hunting and disturbance by farmers in some parts of Europe and North America have caused declines in waterbird populations (Fasola
et al., 2010; Kushlan, 2012). On the contrary, many farmers in Asia have positive attitudes towards nesting waterbirds leading to some
agricultural landscapes supporting globally signicant breeding waterbird populations (Carrasco et al., 2014; Koju et al., 2019; Sundar
et al., 2019). In addition to habitats and human presence, assessments of nest site requirements can be complicated by behaviours such
as nest site delity where waterbirds use the same nest sites for multiple years notwithstanding changes in land use or increase of
human disturbance around nest sites. High site delity is sometimes associated with waterbird population growth that in turn limits
available nest sites (Carrasco et al., 2017). Studies of such sites that are unable to account for waterbird behaviours can provide
potentially incorrect conclusions of nest site associations with landscape scale variables. High site delity was also associated with
S. Kittur and K.S.G. Sundar
Global Ecology and Conservation 30 (2021) e01793
3
improved breeding success, so understanding how sites used repeatedly differ from those that are used only once can signicantly
improve understanding of species conservation (Vergara et al., 2006).
All the waterbird studies referenced above have focused on colonially nesting species that are easily studied due to relative ease in
nding and monitoring colonies, and due to the importance of few large colonies representing signicant populations of several species
(Kushlan, 2012; Parkes et al., 2012; Carrasco et al., 2014; Koju et al., 2019). Territorial waterbird species that build solitary nests on
trees (or “single-nesting waterbirds”) are thinly spread out on landscapes making them much harder to locate. The few existing studies
on single-nesting waterbirds have been carried out in forested and wetland reserves where factors affecting nest locations remained
similar over multiple years, and some species demonstrated strong nest-site delity (Vlachos et al., 2008; Zawardzki et al., 2020;
Fandos et al., 2021; Luzuriage-Neira et al., 2021). In forest reserves maintained for logging, Black Storks (Ciconia nigra) demonstrated
long-term exibility by shifting nest sites from large trees to younger stands in response to logging, suggesting that human activities
forced changes in nesting behaviour (Treinys et al., 2008). Other species like the Oriental White Stork (C. boyciana) nesting in wetland
reserves adapted by nesting on articial structures (electricity poles and pylons) that were further from human presence relative to
trees (Cheng et al., 2020). It is not clear if such exibility in nesting habits is also possible on human-modied landscapes where
landscape scale perturbations due to cropping practices and potentially reduced nest sites may occur in unpredictable ways. It is
possible that single-nesting waterbirds on agricultural landscapes are forced to change nest sites more regularly reecting altering
landscape conditions. Large inter annual variations in nest site choice can likely be indicative of unstable breeding conditions, that can
in turn have serious consequences for species survival. However, exceedingly few studies of breeding ecology of single-nesting wa-
terbirds in agricultural landscapes exist (e.g. Sundar, 2011), and none have as yet explored nest site selection.
The Woolly-necked Stork (C. episcopus) is a single-nesting waterbird found across large swathes of Africa, and south and south-east
Asia that has gained increased scientic attention recently (Gula et al., 2020; Sundar, 2020). Woolly-necked Storks were thought to be
imperilled by agriculture and assumed to require forested reserves (Hancock et al., 1992). However, studies from South Africa, South
Asia, and Myanmar have shown that this species primarily breeds and forages in human-dominated and human-modied landscapes,
including sub-urban settings and farmlands (Sundar, 2006; Thabethe and Downs, 2018; Katuwal et al., 2020; Kittur and Sundar, 2020;
Win et al., 2020; Thabethe et al., 2021; Ghimire et al., in press). In South Africa, this species has taken to supplementary feeding and
shown high nesting propensity on exotic trees and man-made structures in urban and sub-urban areas (Thabethe, 2018). Anecdotal
observations in South Asia have shown Woolly-necked Storks nesting on trees inside cities, on articial structures such as cell phone
towers, and spreading to arid areas following new irrigation canals suggesting that the species adapts to novel man-made conditions
(Choudhary et al., 2013; Singh, 2015; Vaghela et al., 2015; Hasan and Ghimire, 2020; Mehta, 2020). Based on these recent ndings and
Fig. 1. Maps showing location of Haryana state in India (a) and the location of Jhajjar and Rohtak districts in Haryana (b). The spatial distribution
of human habitation (grey polygons in c) and Woolly-necked Stork nests located between 2016 and 2020 (black circles) along with the road routes
taken to survey both districts are shown (c). The spatial distribution of principal landscape features of interest (d: irrigation canals; e: wetlands; f:
tree patches) are also illustrated. Nest locations are randomly staggered by ~1 km to safeguard breeding sites.
S. Kittur and K.S.G. Sundar
Global Ecology and Conservation 30 (2021) e01793
4
an updated population estimate, the conservation status of Woolly-necked Storks has been down listed from ‘Vulnerable’ to ‘Near--
threatened’, but there is still no information on its breeding requirements particularly from South Asia where a signicant proportion
of its global population is found (Gula et al., 2020; Kittur and Sundar, 2020; Sundar, 2020; Win et al., 2020). Information on breeding
requirements is crucial to understand if this species will require tailor-made conservation strategies to allow its long-term survival on
South Asian cereal farmlands. These are characterised by sparse tree cover, extremely high human densities with small farm holdings,
and articially complicated hydrology that supports intensive agriculture with multiple crops harvested each year (Sundar, 2011;
Ricciardi et al., 2018; Koju et al., 2019; Ghimire et al., in press). On such intensively cultivated landscapes, if breeding Woolly-necked
Storks favour relatively rare natural features such as persisting patches of trees and wetlands as cues for locating nests, it would signal a
need to conserve these habitats. Alternatively, if man-made features such as electricity pylons, irrigation canals and trees planted for
agroforestry are providing ecological value for the species, it would indicate the need to support multifunctional traditional agriculture
to ensure long-term conservation of the species.
In this study, we determined correlates of nest locations of Woolly-necked Stork between 2016 and 2020 in the primarily agrarian
and crowded Jhajjar and Rohtak districts of Haryana state in north-central India (Fig. 1). These two districts are ideal for this
investigation for a few reasons.Year-round densities of Woolly-necked Storks in the two districts are among the highest known for the
species (Kittur and Sundar, 2020). Both districts have a network of irrigation canals that are part of the Western Yamuna Canal system
which was renovated in 1335 CE (Jackson, 1999), and have since been substantially increased while also retaining many
community-managed wetlands (Kittur and Sundar, 2020; Fig. 1). Farmers in the two districts have practiced traditional agriculture for
over a century where tree patches and individual scattered trees are included amid croplands as part of multifunctional traditional
agroforestry with trees used for a variety of social, religious, economic and ecological purposes including silviculture, fruits, woodwork
and forage for livestock (Pandey, 2007; see Fig. 1 for tree patches that persist). Finally, farmers and villages have retained wetlands of
varied sizes, alongside human habitations ranging from scattered tiny villages to burgeoning large cities (Fig. 1). These settings allow a
detailed assessment of whether breeding Woolly-necked Storks mimic waterbirds elsewhere in using natural features as cues (tree
patches, wetlands, tall trees) or whether man-made features (agroforestry, electricity pylons, human habitation, irrigation canals)
inuenced nest location.
There has been no previous study of the breeding biology of Woolly-necked Storks in South Asian farmlands. We therefore rst
provide descriptions of several aspects of breeding biology including an assessment of breeding success to allow for comparisons with
data from other landscapes and waterbird species. With breeding success data, we hypothesized that (1) brood size would reect
habitat quality and would therefore increase with proximity to wetlands and decrease with proximity to human habitation. We then
analysed information focussing on variables that past studies have identied as cues used by waterbirds to situate nests in human
modied landscapes. A-piori we hypothesized that (2) Woolly-necked Storks would situate nests near natural habitats (tree patches and
wetlands), away from human presence (habitations) and be neutral to articial sources of water (irrigation canals). Based on these
hypotheses, we constructed spatially explicit habitat suitability models across the two agrarian districts to evaluate their suitability for
nesting storks. We hypothesized that (3) areas of high suitability for breeding Woolly-necked Storks would be restricted to natural
habitat patches. At the scale of the nest tree, we hypothesized that breeding Woolly-necked Storks would (4) preferentially select taller
nest tree species. After analysing all nests collectively, we evaluated if locations of two categories of nests – those with high site delity
(used >1 year) and those on articial structures – were closer to natural habitats and further from human presence mimicking be-
haviours exhibited by other single nesting waterbirds in protected reserves. We hypothesized that (5) these two categories of nests
would reect choice of higher habitat quality by being closer to natural habitats (tree patches, wetlands), further from disturbance
(human habitation), and on taller tree species, relative to nest sites used only once and nests on trees, respectively. Where possible we
developed metrics annually and compared inter-annual results to understand whether Woolly-necked Storks displayed exibility in
locating nest sites or whether cues remained stable across years.
2. Materials and methods
2.1. Study area
Field work was carried out in Jhajjar and Rohtak districts, Haryana, north India across an area measuring 3579 km
2
(Fig. 1). The
two districts were formerly one contiguous district but were separated in 1997 and have human densities of 523 and 608 people/ km
2
respectively (Ofce of the Registrar General and Census Commissioner, India, 2011). Both districts are primarily agrarian with over
80% of land under multi-season cultivation of crops such as corn, millets, rice, and wheat (Singh, 2011). Agriculture is supported by a
vast network of irrigation canals part of the ancient Western Yamuna Command Network set up during the 14th century (Jackson,
1999). Most irrigation canals are unlined at the bottom but have concrete walls. This causes considerable seepage that, along with
percolation from irrigated elds, contributes to ground water recharging resulting in substantial surface waterlogging during the
monsoon and post-monsoon seasons (Singh, 2011). The climate is sub-tropical, semi-arid, and monsoonal with three distinct seasons –
winter (November – February), summer (March – June) and rainfall or monsoon (July – October; Singh, 2011). Annual temperatures
vary between extremes of 2–45 ◦C in the coldest and hottest days respectively, and average annual rainfall, mostly received during the
monsoon, of 590 mm (Singh, 2011). The landscape has sparse tree cover largely as scattered individual trees with few patches. Over a
century of multifunctional agroforestry combined with additional traditional agricultural practices that include trees with specic
utility – both material and spiritual – have helped increase tree cover of both exotic and native tree species primarily along farm
boundaries, roads and canals (Hume, 1889; Umrani and Jain, 2010; Handa et al., 2020; see Fig. 2). Wetlands in Jhajjar and Rohtak
districts are rare (1.2% and 0.38% of the land area respectively) with the majority being seasonal, unprotected, community managed
S. Kittur and K.S.G. Sundar
Global Ecology and Conservation 30 (2021) e01793
5
wetlands (Space Application Centre (ISRO), 2010). There have been exceedingly few ornithological studies in the farmlands of these
two districts (Kittur and Sundar, 2020).
2.2. Field methods
2.2.1. Locating stork nests
Trained eld associates used the existing road network to traverse across both districts covering areas that were accessible in all
seasons (Fig. 1c; see Kittur and Sundar, 2020). Road routes were covered seasonally and continually between 2016 and 2020 using a
motorbike. Woolly-necked Stork nests were located either by directly sighting the nests or using stork behaviour (e.g. carrying nesting
material) to nd nests by following adult birds. The focal species, a representative nest tree monitored during this study and a typical
canal with agroforestry plantation alongside are illustrated in Fig. 2. At each nest, the location of the nest was recorded using a
handheld Global Positioning System and the nest tree species was recorded. Nest fates using robust methods such as repeated nest visits
were not possible due to limited resources. Instead, we provide descriptive metrics of breeding success as the brood sizes of successful
nests that were observed with chicks close to edging.
Fig. 2. The study species, a Woolly-necked Stork (top right), one of the monitored Ficus religiosa nest tree in a crop eld (top left) and a canal with
low water level with a row of Eucalyptus sp. trees planted alongside as part of multifunctional agroforestry (below). Photographs were taken
by authors.
S. Kittur and K.S.G. Sundar
Global Ecology and Conservation 30 (2021) e01793
6
2.2.2. Measuring available tree species
Availability of potential nest trees was enumerated by visiting 244 random points in 2014 that were generated using Arc GIS 10.2
providing one point per 12.2 km
2
and 18.6 km
2
in Jhajjar and Rohtak districts, respectively. The nearest plausible nest tree in each of
the four cardinal directions from each point was visited, identied to species, and height and girth at breast height (GBH) was
measured. A total of 1220 trees were enumerated to assess availability. None of these trees were used for nesting by Woolly-necked
Storks during the study period and were >300 m from monitored nest sites.
2.2.3. Mapping landscape features
Landsat 8 satellite image from 9 October 2018 (U.S. Geological Survey, 2018) covering the entire study area was classied using
unsupervised classication (Isodata clustering) in ERDAS Imagine version 9.1. Pixels in the image measured 30 x 30 m. The image was
classied into 6 land use classes namely agriculture, human habitation, open areas (unutilised for farming or urbanisation), scrublands,
tree patches (including single pixels classied as “trees”) and wetlands. The classication was rened by creating area-of-interest
polygons using onscreen digitization and recoding misclassied areas by checking against reference material such as Google Earth
images and ground-truthing. Classied images had a total accuracy of 92.5%. Irrigation canals were manually digitized from 1:50,000
topographical sheets which had been updated between 2004 and 2006 (Survey of India, 2007).
2.3. Statistical analyses
2.3.1. Descriptive aspects of breeding biology
The breeding season is described with focus on the period of nesting (earliest and latest dates each year when new nests were
located) and edging (latest date of chicks edging). Brood size was documented opportunistically and is presented with the caveat
that this information is biased towards successful nests. However, the compiled information represents the largest ever data set on
breeding success available for this species and is therefore worthy of being presented. As a preliminary examination, correlations
between brood sizes at nests with distances to four spatial variables (irrigation canals, human habitation, tree patches and wetlands)
were examined.
Each year, breeding densities of Woolly-necked Storks along road routes were estimated as the number of nests located/ area
covered, using 150 m on either side of surveyed road as the effective transect width. For this estimate we assumed that the storks could
nest anywhere on the landscape and did not bias estimates to include only supposedly high-quality habitats like tree patches or
wetlands. Since road routes traversed human habitations, agricultural areas, and beside wetlands, computed densities were repre-
sentative of the surveyed landscape. Spatial distribution patterns of stork nests (each year separately and all nests combined) were
evaluated using Ripley’s K-function. Also referred to as the multi-distance spatial cluster analyses, this function summarises spatial
patterns of points (clustered/ dispersed) over a range of distances allowing an evaluation of whether spatial patterns changed with
spatial scale (Haase, 1995). If the average number of neighbours within that distance is higher than the average concentration of points
in the entire study area, points are clustered. We used the multi-distance spatial cluster analysis tool in ArcGIS 10.7 specifying Ripley
edge correction formula as the boundary correction method and minimum enclosing rectangle as the study area method and ran 99
permutations to generate condence envelopes for each run of analysis. The Ripley’s K-function’s output are line graphs showing
expected K or random spatial pattern, condence envelope, and observed K or spatial pattern at varying distances. The pattern is
clustered when observed K >expected K and dispersed when expected K values >observed K values at a particular distance. Observed
K values that are outside of the condence envelope have patterns (clustered or dispersed) that are statistically signicant at the 95%
level. We developed graphs for each year and for all nest sites combined.
2.3.2. Modelling nest locations against land use variables
Four spatial layers were generated, each representing distances (m) to the primary variables of interest: irrigation canals, human
habitation, tree patches, and wetlands. These layers were generated using the Euclidian distance tool in Arc GIS 10.7.1. For each year
of eld work, we used spatial logistic regressions using function “glm” in statistical platform R to understand if distances to these
variables varied between nest sites and randomly generated points. Random points equalling the number of nest locations each year
were generated on surveyed tracks/road routes where nest sites were not present. The road routes within the study area were used as
the constraining feature to generate random points after erasing a buffer of 1 km around each nest location. All logistic regression
models were run after splitting the data into training (70%) and test (30%) datasets to evaluate model performance. We developed
spatial logistic regressions rst for each year to understand cues used by Woolly-necked Storks to locate nests each year, and to un-
derstand if storks used the same cues across years. We also combined all nest sites across years to measure the “full” strength of re-
lationships during the study period. We combined data sets in two ways. We rst used nests with site delity (nest sites used >1 year)
only once and then also ran a model where nest sites used >1 year were included multiple times matching the number of years they
were observed to be reused. Model output when we used each nest site only once were not representative of models developed for each
year. We therefore only included metrics and the map developed with the model that included nest sites multiple times when they were
used more than once (N =298). We diagnosed model utility by computing model sensitivity (proportion of points evaluated as positive
correctly) and model specicity (proportion of points evaluated as negative correctly) after using the test dataset to make predictions.
We plotted the receiver operating characteristics (ROC) curve and the area under ROC (AUROC) using package “InformationValue”
(Prabhakaran, 2016). If selected pairs of observations (one with an outcome of 0, or point with no nest, and one with 1, or location of a
nest) are drawn at random, AUROC is the probability with which the model correctly ranks these pairs of observations. The AUROC
varies from 0.5 (model with no discrimination ability) to 1 (model with perfect discrimination ability). Models with high
S. Kittur and K.S.G. Sundar
Global Ecology and Conservation 30 (2021) e01793
7
discrimination capability will have high sensitivity and specicity scores simultaneously, and ROC-curves of these models will be
consistently higher than the 45◦diagonal line. We computed McFadden’s r
2
for each model, which represents the likelihood of each
observation correctly predicting the outcome of the model. Models that explain all the variation in the outcome will have McFadden’s
r
2
=1, but this outcome is almost never possible in natural settings given the inherent complexity of conditions. The low probability of
nding variables that would lead to perfect model ts in studies of novel situations further reduces the possibility of getting high values
of McFadden’s r
2
. Rather than using this metric as representing model strength, we use McFadden’s r
2
to identify potential model
latency. Partial dependence plots were developed to document directionality of the effects of variables (positive or negative) using
package “pdp” (Greenwell, 2018). We did not incorporate or evaluate potential interaction effects between variables. We used the
intercept and beta-estimate coefcients from the logistic regression models using all variables to develop maps of nest location
suitability. We extracted total area in each of four classes of suitability (0 – 0.25; 0.25 – 0.5; 0.5 – 0.75; >0.75) to assess the extent of
inter-annual variations in suitable habitat in Jhajjar and Rohtak for Woolly-necked Storks to situate nests. We improved reliability of
models in three ways. First, we incorporated true absences alongside presence locations, which is critical for modelling robust habitat
suitability at smaller spatial scales (Brotons et al., 2004). Second, we avoided using general climatic variables instead using high
resolution biophysical variables which are often more important determinants of species distribution (Manzoor et al., 2018). Finally,
we avoided extrapolating our ndings to outside of the focal study area since relationships with biophysical variables can vary with
location (Brotons et al., 2004; Manzoor et al., 2018). The methods we use for modelling habitat suitability and model diagnostics are
widely used and well established in ecology (Fielding and Bell, 1997; Brotons et al., 2004).
2.3.3. Nest tree preference
To understand whether Woolly-necked Storks exercised selection of nest tree species, we employed the use-availability framework
(Manly et al., 2004). We used function ‘widesI’ in R-package “adehabitat” (Calenge, 2006) which does not require data from indi-
vidually marked animals. The algorithm computed the Manly selectivity measure to test selection of nest tree species using two scales.
The rst scale was for the overall data set estimated using log-likelihood
χ
2
(or the ‘Khi2L’ statistic of “adehabitat”) testing the hy-
pothesis that all available nest tree species were used randomly. The second scale of selection was at the level of the tree species and
used pair-wise Bonferroni tests with use/ available proportions to compute selection ratios. These ratios provided a statistical
assessment of whether each nest tree species was preferred (used more relative to availability), avoided (used less relative to avail-
ability but had some nests) or used in proportion to availability (Manly et al., 2004). The function also provided scaled Manly’s
selectivity ratios (B
i
) for each nest tree species summing to 1 across all nest tree species. This selectivity ratio is interpreted as being the
estimated probability that a tree of a particular species would be the next one selected by storks to nest on if it were possible to make
individual trees of all species equally available (Manly et al., 2004). The value of the ratio would not therefore indicate preference or
avoidance but would help identify species that were important, taking into account the differential availability of the various tree
species. For these analyses, we combined tree species without any nests into one class and excluded nests on articial structures. We
rst analysed the data year-wise to assess whether storks exercised choices similarly each year, and then analysed the data combining
observations across all years since annual variations were minimal (see Results).
2.3.4. Site delity and nests on articial structures
We asked if sites with nest delity (sites used >1 year) were different from sites used only once by contrasting distances to each of
the four variables. We performed non-parametric permutation tests with package “coin” in R (Hothorn et al., 2021) to compare
distances between nests and the four variables of interest. Using non-parametric tests allowed us to use data sets that did not
Table 1
Descriptive metrics and summaries of distances measured to four primary variables from nests of Woolly-necked Storks in Jhajjar and Rohtak districts,
Haryana. Combined sites include all 298 nests monitored during the study period (2016 – 2020). Distance values are average ±SD (range).
2016 2017 2018 2019 2020 Combined
Descriptive metrics
No. nests 67 73 49 53 56 298
Nests km
-2
0.06 0.06 0.04 0.04 0.05
–
% nest in
habitation
1.5 0 4.1 1.9 1.8 1.68
% nest on pylons 8.96 13.73 8.16 15.09 8.93 10.07
Brood size
No. nests 13 12 8 6 3 42
Average±SD 2.5 ±2.5 3.2 ±0.9 3.5 ±0.8 3.2 ±0.8 3.3 ±0.6 3.1 ±0.9
Median 3 3 4 3 3 3
Distances (m) of nests to focal landscape scale variables
Irrigation canals 179 ±207 (0–924) 143 ±181 (0–932) 162 ±176 (0–845) 175 ±187 (0–845) 161 ±246
(0–1110)
163 ±200
(0–1110)
Habitation 757 ±447 (0–1615) 762 ±435
(42–1720)
721 ±496
(0–2115)
856 ±524 (0–2098) 775 ±498
(0–1728)
773 ±475
(0–2115)
Tree patches 1098 ±574
(120–2892)
1176 ±578
(153–2892)
1071 ±588
(42–2647)
1039 ±490
(134–2089)
971 ±572
(0–2290)
1078 ±564
(0–2892)
Wetlands 655 ±399 (0–1888) 658 ±382 (0–1917) 656 ±387
(30–1557)
594 ±345 (0–1557) 631 ±387
(0–1684)
641 ±380
(0–1917)
S. Kittur and K.S.G. Sundar
Global Ecology and Conservation 30 (2021) e01793
8
necessarily conform to tests of uniformity.
We tested whether articial structures (electricity pylons) with stork nests varied in their distance to each of the four variables
relative to locations where nests were placed on trees. For these comparisons, we again used non-parametric permutation tests. Nests
on pylons were rare relative to nests on trees each year. To correct for the unbalanced dataset, we specied type III sums of squares
which adjusts the sums of squares to estimate what they might have been had the data been balanced (Milliken and Johnson, 2009).
3. Results
3.1. Woolly-necked Stork breeding biology
Each year, Woolly-necked Storks nested between the rst week of May and last week of October. Density of nests varied little
annually ranging between 0.04 and 0.06 nests km
-2
(Table 1). Each year a small proportion of nests (8.16 – 15.09%) were located on
articial structures, all of them on electricity pylons (Table 1). Also, each year few nests were found within areas of human habitation
(0 – 4.1%; Table 1). Average brood size of 42 successful nests was 3.1 ±0.9 SD (annual average brood size range: 2.5 – 3.5) with a
median of three chicks most years (Table 1). Broods of three and four were the most frequent and two broods had ve chicks each
(Fig. A.1). Univariate correlations between brood size and variables showed statistically signicant smaller brood sizes in nests closer
to habitation (Pearson’s product-moment correlation, r =0.34, P =0.03) and wetlands (r =0.32, P =0.04). Brood sizes did not show
a statistically signicant correlation with distance to tree patches (r
2
=0.25, P =0.11), and were negatively but only weakly correlated
with distance from canals (r
2
= − 0.07, P =0.68). Taking all nests into account, nests were located furthest to tree patches and nearest
to irrigation canals and these trends did not alter annually (Table 1).
Ripley’s K-functions indicated that nests were randomly distributed only at the smallest spatial scales but were otherwise signif-
icantly clumped, with the degree of clumping increasing as the spatial scale of assessment increased (Fig. A.2). Spatial patterns of nest
locations were largely similar across years, and for the full data set.
3.2. Landscape variables inuencing nest locations
Nests were much closer to irrigation canals and further from human habitation relative to random locations (Tables 1, 2; Fig. 3).
Notwithstanding the number of nests located each year, distance to irrigation canals and habitation were the only statistically sig-
nicant variables every year and for the combined data set (Table 2). The directionality of these relationships stayed the same each
year and for the full model, though partial dependence plots showed that directionality of relationships with the other two variables
changed in some years (Fig. A.3). Models showed relatively high sensitivity but varying specicity each year. Area under the ROC for
the combined data set was relatively high at 0.72 (Table 2; Fig. A.4). Annual values of AUROC however showed some variation
(AUROC range =0.61–0.88; Table 2; Fig. A.4). Despite two variables showing statistical signicance in explaining observed vari-
ability, MacFadden’s r
2
suggested considerable model latency and that the four variables used were explaining only a small amount of
the inherent variability in nest location by Woolly-necked Storks (Table 2). Habitat suitability modelling showed considerable swathes
of the northern and central parts of Jhajjar and Rohtak districts to be moderately (suitability >0.5) and highly suitable (suitability >
0.75) for Woolly-necked Storks to situate nests (Fig. 4; Table A.1). Total area under each of the four classes of habitat suitability varied
substantially among yearly models (
χ
2
=1851, d.f. =16, P <0.001; Table A.1). Modelled area with the highest suitability was the
most during 2018 (37%; Fig. 4; Table A.1).
3.3. Nest tree species choice
Sampling for availability measured 1220 trees of 46 species with the commonest species (29.5% of all trees measured) being the
native Dalbergia sissoo, a tree with tough wood coveted for making furniture and agricultural yokes. The next two most abundant trees
Table 2
Results of logistic regressions used to measure effects of distances to four primary variables from nests of Woolly-necked Storks versus random lo-
cations in Jhajjar and Rohtak districts, Haryana. Values are beta-estimates for each variable estimated using full models that included all four
variables. Symbols indicate levels of statistical signicance (*
–
<0.05; **
–
<0.01; ***
–
<0.001). Four measures of model diagnostics are
presented to enable model evaluation.
2016 2017 2018 2019 2020 Combined
Metrics from logistic regression
Intercept -0.2138 -0.3287 -0.6391 0.2078 0.1875 -0.0309
Irrigation canals -0.0017 -0.0048 ** -0.0038 * -0.0024 * -0.0020 * -0.0021 ***
Habitation 0.0012 * 0.0014 * 0.0028 ** 0.0013 * 0.0017 ** 0.0009 ***
Tree patches -0.0003 0.0004 -0.0002 0.0003 -0.0005 0.0002
Wetlands 0.0004 0.0001 0.0009 -0.0013 -0.0002 -0.0004
Model evaluation
Sensitivity 0.722 0.714 0.816 0.765 0.706 0.843
Specicity 0.476 0.667 0.6 0.75 0.526 0.58
AUROC 0.605 0.693 0.813 0.886 0.686 0.716
McFadden’s r
2
0.103 0.226 0.288 0.161 0.166 0.101
S. Kittur and K.S.G. Sundar
Global Ecology and Conservation 30 (2021) e01793
9
Fig. 3. Ridgeline plots showing differences in distances to irrigation canals (a) and human habitation (b) of Woolly-necked Stork nest locations
(dark grey) versus random locations (light grey) in Jhajjar and Rohtak districts, Haryana. Distances to other land uses were not statistically
signicantly different between nest and random locations and are not shown.
Fig. 4. Habitat suitability of Rohtak and Jhajjar districts, Haryana, for Woolly-necked Storks to situate nests. Variables used were distances to
irrigation canals, human habitation, tree patches, and wetlands. Annual sample sizes and strengths of associations with individual variables are
detailed in Table 1.
S. Kittur and K.S.G. Sundar
Global Ecology and Conservation 30 (2021) e01793
10
were also both native, Acacia nilotica (11%) and a tree popular for roadside plantations, Azadirachta indica (11%). Woolly-necked Stork
nests, however, were mostly on D. sissoo (30% of 152 unique nest sites that were on trees), Ficus religiosa (a native wild species regarded
as holy and was rare on the landscape – see Fig. 2; 29% of nest sites), and Eucalyptus sp. (an exotic species favoured for agroforestry;
25% of nest sites; Figs. 2 and 5). Nests were found on 10 tree species suggesting that, at the landscape scale, Woolly-necked Storks
displayed strong non-random use of available trees (
χ
2
values in Table 3). D. sissoo, Mangifera indica, Mitragyna parviora and Tectona
grandis were used in proportion to their availability while F. religiosa and Eucalyptus sp. were used much higher relative to availability.
Acacia nilotica, A. indica, Syzygium cumini and 37 additional species were rarely used for nesting (Table 3; Fig. 5). F. benaghalensis had
only one nest. Patterns of nest tree species choice were similar each year (Table 3). M. parviora and T. grandis had the largest scaled
Manly’s selection ratios suggesting that Woolly-necked Storks would likely use these two tree species much more to situate nests if the
trees were more abundant (Table 3).
3.4. Nest delity and sites on pylons
Woolly-necked Storks used 44.5% of 166 unique nest sites more than once between 2016 and 2020. Sites with nest delity were
located signicantly further from wetlands relative to sites used only once (Table A.2a). Nests on pylons were reused far more than
nests on trees though the nests reused for three and four years were on trees (Fig. 6a). Nests on trees were located signicantly closer to
irrigation canals relative to nests on pylons but had similar distance to other variables (Table A.2b). Woolly-necked Storks renested on
all three primary nest tree species similarly (Fig. 6b). Two nests that were used throughout the study period were on D. sissoo (Fig. 6b).
4. Discussion
We evaluated the value of two agrarian north Indian districts for nesting Woolly-necked Storks and discovered several aspects novel
to the broad subject of conserving biodiversity in agricultural landscapes. Contrary to ndings in protected reserves regarding nesting
behaviour of single nesting waterbirds, the Woolly-necked Stork breeding population in the intensively cultivated and crowded dis-
tricts of Jhajjar and Rohtak was surprisingly high. Additionally, storks situated their nests using cues related to agriculture and were
neutral to persisting natural habitats. We have uncovered a landscape where current agricultural practices retain many elements of
ancient traditional agriculture such as planting and retaining many tree species alongside crops, with an increase in canal-fed irrigated
crops. These are supporting and benetting a waterbird species that, until recently, was incorrectly assumed to be declining due to
agriculture.
4.1. Woolly-necked Stork breeding biology
Most of the aspects of breeding biology we present here are novel since most of the past published observations on this species have
been anecdotes. We recorded Woolly-necked Stork nesting between May and October which is a much wider nesting season than the
June to September season reported earlier for this species in north India (Ali and Ripley, 2001). Woolly-necked Stork nest density
remained similar year-to-year suggesting that breeding birds are resident and territorial. This nding adds to existing information
showing Jhajjar and Rohtak districts supports one of the largest known populations of resident Woolly-necked Storks (Kittur and
Sundar, 2020). Estimated densities also suggest that agricultural areas in south Asia hold the potential to support tens of thousands of
Woolly-necked Stork pairs at the least. It is not immediately clear why nests followed a clumped distribution, though visual inspection
of maps show that locations of habitations and irrigation canals – two variables that inuenced nest locations strongly – were not
uniformly distributed on the landscape (Fig. 1d,f). These variables may have affected observed patterns of nest distribution.
Fig. 5. Primary nest trees used by Woolly-necked Storks to nest on in Jhajjar and Rohtak districts, Haryana between 2016 and 2020. Three nest tree
species on which storks situated most of their nests are illustrated to allow comparisons in height and crown shapes, and include placement of a
typical nest. Tree heights represented are average values measured for available trees at random locations. Bar-graphs show comparisons of the
availability of the three tree species on the landscape (black) and their use for nests (152 unique nest sites) by storks (white). Selection metrics are
provided in Table 3. (Graphics were created by the authors.).
S. Kittur and K.S.G. Sundar
Global Ecology and Conservation 30 (2021) e01793
11
Woolly-necked Storks rarely used articial structures to situate nests (Table 1). Pylons in Haryana were usually located away from
human habitation, and it would not have been unreasonable to nd more nests on this substrate. Anecdotal observations of Woolly-
necked Storks using human-made structures such as cell phone towers have been suspected to be related to reduction of nest trees and
to reduced human disturbance (Vaghela et al., 2015; Hasan and Ghimire, 2020). Our documentation does not provide evidence of
either supposition since Jhajjar and Rohtak did not have a paucity of nest trees, and nests on pylons were not further from human
Table 3
Scaled Manly’s selection ratios (B
i
) for tree species used for nesting by Woolly-necked Storks in Jhajjar and Rohtak districts, Haryana. Tree species that
had at least three nests during the study period are listed separately and the rest have been included in “Others”.
χ
2
provides results for the test of the
hypothesis that use of all habitat classes was random. Asterisks indicate levels of statistical signicance (* <0.05;
**
<0.01;
***
<0.005) and symbols
indicate whether tree species were preferred (+) or avoided (-) relative to their availability. Tree species that were used in proportion to their
availability do not have asterisks or symbols, and species that were not used in a particular year do not have any values. Scaled B
i
ratios each year sum
to 1. Values in bold highlight tree species that were preferred or avoided by storks. For example, the results for 2016 shows that storks nested on ve of
the listed nest trees, MIPA was the highest ranked nest tree, three tree species were preferred (AZIN; EUCA; FIRE), and two tree species were used in
proportion to their availability (DASI; MIPA) for nesting. (Trees: ACNI – Acacia nilotica; AZIN – Azadirachta indica; DASI – Dalbergia sissoo; EUCA –
Eucalyptus sp.; FIRE – Ficus religiosa; MAIN – Mangifera indica; MIPA – Mitragyna parviora; SYCU – Syzhygium cumini; TEGR – Tectona grandis.).
Trees 2016 2017 2018 2019 2020 Combined
χ
2
117.9 * ** 127.6 * ** 101.3 * ** 97.7 * ** 114.3 * ** 305.3 * **
ACNI 0.002 * ** (-) 0.008 0.004 * ** (-)
AZIN 0.006 * ** (-) 0.004 * ** (-) 0.005 * ** (-) 0.002 * ** (-) 0.003 * ** (-) 0.004 * ** (-)
DASI 0.049 0.029 0.025 0.012 0.012 0.023
EUCA 0.121 * (þ) 0.128 * ** (þ) 0.064 * (þ) 0.033 * (þ) 0.067 * * (þ) 0.087 * ** (þ)
FIRE 0.442 * ** (þ) 0.252 * ** (þ) 0.287 * ** (þ) 0.092 * ** (þ) 0.149 * ** (þ) 0.21 * ** (þ)
MAIN 0.061 0.033 0.043 0.062
MIPA 0.378 0.503 0.605 0.275 0.356 0.257
SYCU 0.021 0.013 0.008 0.007 * * (-)
TEGR 0.55 0.356 0.343
Others 0.004 * ** (-) 0.002 * ** (-) 0.001 * ** (-) 0.001 * ** (-) 0.001 * ** (-) 0.002 * ** (-)
Fig. 6. Woolly-necked Stork nest site delity between 2016 and 2020 at Jhajjar and Rohtak districts, Haryana. Bar graphs show comparisons of the
number of times sites were used when nests were located on pylons or trees (a), and when nests were located on different tree species (b). Numbers
in parenthesis indicate number of nest sites. In (b), “Other” refers to nests found on tree species (31 species) other than the three on which most nests
were observed and excludes nests on pylons. (Woolly-necked Stork sketch by Saniya Chaplod.).
S. Kittur and K.S.G. Sundar
Global Ecology and Conservation 30 (2021) e01793
12
habitation relative to nests on trees (Table A.2b). Woolly-necked Storks in Haryana nested on man-made structures to a similar extent
as in peri-urban and urban areas in South Africa (8.43% of 166 unique nest sites in Haryana versus 10% of 30 nests in KwaZulu-Natal;
Thabethe, 2018). This is suggestive of the species preferring trees as nesting substrates over articial structures.
Despite the preliminary nature of the data on breeding success, correlations suggest that Woolly-necked Stork nest locations may
confer advantages in terms of brood size. Brood sizes in this study are the highest known for any single nesting waterbird species and
are comparable or higher than those recorded for colonially nesting species notwithstanding the kind of landscapes waterbirds nests in
(Tryjanowski et al., 2006; Vergara et al., 2006; Koju et al., 2019; Sundar et al., 2019; Luzuriage-Neira et al., 2021). A major caveat of
this nding is that we present only observations of successful nests, and it is possible that more robust consideration of all nests may
yield lower brood sizes. Larger brood sizes were in nests that were located further away from human habitation which is suggestive of
negative impacts of human disturbance on breeding birds. It is likely that foraging quality was higher away from concretised urban
areas, especially since observations in South Africa showed Woolly-necked Storks largely provisioning chicks with amphibians
(Thabethe, 2018). The correlation between brood size and distance to habitation matches the results of habitat suitability modelling
for nest locations that showed nests to be located further from habitation. Nests with larger broods were also further from wetlands.
Observations in neighbouring districts have shown extensive illegal conversions of wetlands to sh farms and reduced usage by some
waterbirds, which is suggestive of deterioration of ecological conditions (Sundar et al., 2015). It is possible that conditions at wetlands
are deteriorating in Jhajjar and Rohtak as well and was reected in the choice of Woolly-necked Stork nest sites. A specic assessment
to evaluate the condition of wetlands at large across the two districts will be necessary to clarify this possibility. Brood sizes showed
negative correlations with distance from canals. While the relationship was statistically not signicant, it strengthens the suspicion that
storks select nest locations that maximise provisioning potential. The use of the same cues to situate nests by storks year after year
suggests relatively high stability of conditions for breeding on the landscape. Rainfall intensity and patterns, hydro periodicity in
canals, the main crops on the landscape, and stork territoriality are potential variables that additionally affected Woolly-necked Stork
nest site choice and brood sizes each year. The large brood sizes of Woolly-necked Storks in the largely agrarian districts of Jhajjar and
Rohtak indicate a healthy system supporting a signicant breeding population of a large waterbird.
4.2. Habitat suitability and important variables
Contrary to our hypotheses, Woolly-necked Storks did not situate nests close to wetlands or tree patches on the landscape (Table 2).
Situating nests away from discrete tree patches suggests that Woolly-necked Storks were not averse to nesting on single, scattered trees
or on small lines of trees that were not detected on satellite imageries with the resolution we used (see Fig. 2). Storks did not entirely
avoid habitation but showed a signicant aversion to habitation by consistently nesting away from human settlements in line with our
hypothesis (see Table 1). This habit of nesting within areas of human habitation was likely linked, at least in part, to site delity and
were potentially pairs whose nest sites became part of recent expansions of smaller towns (personal observations). Woolly-necked
Storks in Haryana used nest site selection cues that were different from those shown by the same species in South Africa which had
a strong preference for nesting on trees inside residential yards close to swimming pools (Thabethe, 2018). These detailed studies of
Woolly-necked Storks in Asia and Africa suggest that this species does not show a proclivity to natural wetlands.
The major novel nding of this study was the use of the extensive irrigation canal network as cues by Woolly-necked Storks to
situate nests. This habit appears to be over a century old as noted by British naturalists (Hume, 1889). During our study, Woolly-necked
Storks did not preferentially select trees close to wetlands that are commonly regarded as superior foraging habitats. Most wetlands in
Jhajjar and Rohtak were community waterbodies that experienced considerable use by people and livestock throughout the year
(personal observations). On the other hand, irrigation canals and the small wetlands formed alongside canals due to leakages expe-
rienced much less use and this reduced human presence may have attracted storks to canals. Additionally, wetlands were strongly
seasonal while canals experienced longer hydroperiods relative to most wetlands (personal observations). It was also common to
undertake multifunctional agroforestry along irrigation canals (see Fig. 2), a traditional practice that has been around for centuries.
The combination of reduced human presence near canals, increased year-long availability of water in canals and availability of nest
trees along canals may be responsible for Woolly-necked Storks locating their nests near irrigation canals. Despite statistical signi-
cance of two variables, model diagnostics pointed to considerable latency in results suggesting that the full model used explained only
a little of the overall variation inherent in the system. It is likely that additional variables such as food availability also contribute to
storks’ choice of nest location, especially since Woolly-necked Stork diet may not be restricted to species found entirely in wetlands. In
South Africa, Woolly-necked Storks mostly provisioned chicks with amphibians (Thabethe, 2018). In Haryana, amphibians may be
easier to catch in shallow pools, croplands, and canals with low water levels, relative to in deeper wetlands. Future work should focus
on identifying stork diet and the contribution of different habitats in contributing different dietary items. Such work may provide more
insights into why storks are locating nests as observed in this study. Studies over much longer time frames will be needed, alongside
individually marking breeding birds, to understand patterns of site delity of breeding Woolly-necked Storks. Our analyses, and the
study from South Africa (Thabethe, 2018), shows that high resolution annual data and variables relating to individual landscapes are
needed to understand the factors inuencing the species’ nest site selection.
Area modelled as the highest level of suitability varied considerably each year (10 – 37%; Table A.1). These proportions were,
however, much larger than that estimated for the model with all nests included (5%) suggesting that analyses that combine nests across
years or sites with varied conditions require careful interpretation to avoid misunderstanding the species’ requirements.
S. Kittur and K.S.G. Sundar
Global Ecology and Conservation 30 (2021) e01793
13
4.3. Choice of nest trees
In our study we avoided disturbing nests and did not measure nest trees and are unable to provide an understanding of whether
storks preferentially nested on trees of specic size classes. Waterbirds, including Woolly-necked Storks in KwaZulu – Natal, are known
to select the largest trees even among preferentially used species, and it is highly likely that Woolly-necked Storks in Haryana also
followed the same pattern (Treinys et al., 2008; Thabethe, 2018; Koju et al., 2019). Woolly-necked Storks in South Africa situated most
of their nests on trees in residential areas, using the tallest tree species that were also mostly exotic species, and included Eucalyptus sp.
(Thabethe, 2018). In Haryana, that Woolly-necked Storks largely nested on a medium-sized tree is likely an indication of the absence of
directed persecution by farmers and appears to be another example of South Asian farmers practising attitudes and farming styles that
favour waterbirds. One of the trees that was used much more relative to its availability was F. religiosa – a tree species that is considered
holy by Indian farmers and is not removed despite its considerable size and canopy cover that shades crops and potentially reduces
crop productivity (see Fig. 2). The Woolly-necked Storks’ habits of using D. sissoo and F. religiosa trees amid cultivation and along
canals for nesting in Haryana and other parts of north India is not recent. Observations from the 1800s include the following statements
(parentheses are ours): “sheeshum (D. sissoo) being their favourite (nest tree)”, “several nests on peepul (F. religiosa) mostly in the
neighbourhood of canals”, and “nest on a banyan tree (F. benghalensis) in a grove” (Hume, 1889). These observations, though anec-
dotal, underscore the long-standing value of traditional multifunctional agriculture for breeding Woolly-necked Storks in north India.
This combination of directed persecution of the birds being absent and traditional agricultural practices that include trees alongside
crops are present in agricultural areas across South Asia, but their value for waterbirds is being documented only relatively recently. So
far there is documented evidence for this combination of factors to be sustaining the largest known breeding population of Lesser
Adjutant Storks (Leptoptilos javanicus) – a globally Vulnerable waterbird species (Koju et al., 2019; Sundar et al., 2019) in lowland
Nepal, and the largest known breeding population of Woolly-necked Storks (this study). The majority of investigations on traditional
multifunctional agroforestry in South Asia have focused on economic and material benets of trees retained amid croplands (e.g.
Dhyani et al., 2009), with little work on ecological benets of trees. This lacuna requires correction and improving understanding of
the value of agroforestry to species conservation can assist to reduce assumptions regarding the impacts of small holder tropical
agriculture on biodiversity conservation.
4.4. Nest site delity
There is unfortunately no comparable data set with which to infer whether our observed nest site delity of ~45% is low or high for
single-nesting storks using farmlands. Reuse of nest sites is common in storks (Hancock et al., 1992; Tryjanowski et al., 2006; Vergara
et al., 2006; Fandos et al., 2021), and has been observed in Woolly-necked Storks in South Africa (Thabethe, 2018) and India (Pur-
basha, 2017). This study provides conrmation that reuse of nesting sites by Woolly-necked Storks in South Asia is part of a general
pattern observed in storks. Our data set suggests that nests on pylons were reused to a larger degree relative to nests on trees (Fig. 6a).
Since most nests were on trees favoured in multifunctional agroforestry, it is possible that some trees each year were disturbed by
forestry activities. However, we also observed that storks have similar patterns of nest site reuse on trees regardless of the tree species
or their use by farmers (Fig. 6b). These two ndings suggest that agroforestry practices alone cannot explain higher levels of reuse of
nest sites on pylons.
Nest sites on pylons were, on average, further from canals relative to nests on trees – contrary to the overall ndings in the study
(Table A.2a). Woolly-necked Stork behaviour in Haryana also did not match ndings in protected reserves elsewhere where Oriental
White Storks took to nesting on articial structures to avoid human presence (Cheng et al., 2020). However, reused Woolly-necked
Stork nest sites in Haryana were further from wetlands relative to sites used only once (Fig. 3). Long-term studies on other stork
species have shown older birds with more experience reusing nests for longer durations (Vergara et al., 2006). If older Woolly-necked
Storks are also the ones reusing nest sites, then the large number of sites used once would suggest that much of the breeding population
of storks in Jhajjar and Rohtak is relatively young, and breeding pairs are still working on nding the best nest sites. It would also
explain why nest sites used only once were closer to wetlands – younger and less experienced pairs could be attracted to wetlands
initially before learning to select sites closer to canals to reuse for multiple years.
4.5. Implications of the study
Multiple lines of evidence underscore the high quality of the ancient agrarian areas of Haryana for breeding Woolly-necked Storks,
and historical literature provide anecdotes that show this situation to be existing since at least the 1800 s. There is no previous study
where waterbird nest location was inuenced by agricultural aspects on landscapes that still retained natural habitats. This indicates
that farming methods and cropping patterns in Haryana are conducive to maintain breeding stork populations, similar to recent
observations from lowland Nepal where farmers follow similar traditional farming (Koju et al., 2019; Sundar et al., 2019). In addition
to conducive farming methods and suitable crops, the critical ingredient for such multifunctional agricultural landscapes appears to be
favourable farmer attitudes that, at the minimum, do not actively persecute nesting waterbirds. Additional aspects common to lo-
cations where storks are breeding in small holder farmlands is traditional agriculture that includes trees amid crop lands. Trees were
grown both for material use (e.g. D. sissoo for furniture and making yokes) and for spiritual purposes (e.g. Ficus species that are
regarded as holy in Hinduism). This combination of factors appears to be invaluable to conserve biodiversity and conservationists
should develop strategies that encourage these factors across agricultural landscapes. These factors are, however, unlikely to be
transferable to other areas including to the developed global north where large-scale mechanisation requires removal of scattered trees
S. Kittur and K.S.G. Sundar
Global Ecology and Conservation 30 (2021) e01793
14
on agricultural landscapes, and where locals hunt waterbirds or regard waterbirds nesting outside of protected reserves as a nuisance.
The absence of the spiritual element in farmers of the developed countries is a critical lacuna, and conservation efforts in countries such
as India require to embrace these important differences between regions. It is essential that different agricultural regions be evaluated
separately without imposing ndings from one area on the other, or assuming that there are silver-bullet strategies that work with
similar efciency in areas with disparate farmer habits. The continuance and expansion of traditional agriculture that is benetting
wild species, such as in Haryana, should be supported. There is an urgent need to avoid imposing protectionist paradigms at such
locations, instead recognizing the value of existing farmer-nature connections which are easily degraded with the unsustainable but
widely prevalent interventions such as payment-based conservation (Fischer et al., 2012).
In this study, we take a strictly ecological approach biased towards deciphering the needs of breeding Woolly-necked Storks. Model
diagnostics have showcased that these measurements are insufcient to fully explain the observations. We suggest that a multidis-
ciplinary approach that includes social, economic, and agricultural dimensions may uncover a fuller set of mechanisms that favour
waterbird breeding on agricultural landscapes. The number of ecological investigations in agricultural landscapes outside of developed
countries are still sparse and a wider understanding of how different agricultural settings could enable biodiversity conservation still
eludes us. Nonetheless, our discovery of an agricultural patchwork sustaining the largest known breeding population of a poorly
studied waterbird is signicant, as is the observation that its success is driven by features introduced to benet farmers. Natural
hydrological ows are increasingly being harnessed for agriculture and a growing number of locations are witnessing declines in
natural wetland functions and declines in breeding waterbirds (Brandis et al., 2018; Bino et al., 2020). Our work demonstrates that
increasing irrigation can, under certain settings, improve breeding conditions for waterbirds. Learning how to harness this advantage
alongside cultivation can confer signicant long-term conservation benets with minimal additional expenditure. There are several
other waterbird species in Asia and Africa that are currently assessed as species of global conservation concern. The conservation status
of many of these species is based largely on the assumption that farmlands are detrimental to their wellbeing despite an absence of
surveys in agricultural areas (Gula, 2020; Sundar, 2020). Accumulating evidence shows that rather than make such unsubstantiated
assumptions, we require to identify waterbirds whose conservation status cannot be assessed reasonably with existing information and
identify agricultural landscapes that hold the potential to be multifunctional and could be supporting these waterbird species. Like in
Jhajjar and Rohtak, it seems entirely plausible that additional mechanisms that will be novel to scientic literature exist outside of
strictly protected reserves and support waterbird populations. The density of irrigation canals in Jhajjar and Rohtak is perhaps un-
matched in South Asia. However, irrigation canals are a common feature of most South Asian farmlands, as are other features such as
tanks and reservoirs. It is highly unlikely that these existing features are without positive effect on waterbirds. Planning careful work to
suit each landscape is needed to compile diverse and potentially novel mechanisms to enrich existing toolkits towards enhancing
biodiversity conservation amid agriculture. A good rst step towards building such toolkits will be to stop caricaturing crowded small
holder agricultural areas of the global south as being uniformly detrimental to all waterbirds.
Funding
The Bryan Guinness Foundation and National Geographic Foundation.
Declaration of Competing Interest
The authors declare that they have no known competing nancial interests or personal relationships that could have appeared to
inuence the work reported in this paper.
Acknowledgements
For continued support we thank Lata Kittur and Sanjay Prasad. Administrative support was provided by the International Crane
Foundation and the Nature Conservation Foundation. We gratefully acknowledge A.S. Chauhan, R. Ahlawat and D.S. Dalal for con-
ducting eld work towards this long-term monitoring project. Comments and discussions on a previous draft by J.D.A. Grant, J. Gula
and S. Subramanya greatly improved the manuscript, and Saniya Chaplod kindly provided her sketch for use in this work. A previous
draft benetted from the thoughts of three anonymous reviewers.
Appendix A. Supporting information
Supplementary data associated with this article can be found in the online version at doi:10.1016/j.gecco.2021.e01793.
References
Ali, S., Ripley, S.D., 2001. Handbook of the Birds of India and Pakistan, Vol. 1. Oxford University Press, New Delhi, India.
Bino, G., Brandis, K.J., Kingsford, R.T., Porter, J.L., 2020. Waterbird synchrony across Australia’s highly variable dryland rivers – risks and opportunities for
conservation. Biol. Conserv. 243, 108497 https://doi.org/10.1016/j.biocon.2020.108497.
S. Kittur and K.S.G. Sundar
Global Ecology and Conservation 30 (2021) e01793
15
Brandis, K.J., Bino, G., Spencer, J.A., Ramp, D., Kingsford, R.T., 2018. Decline in colonial waterbird breeding highlights loss of Ramsar wetland function. Biol. Cons.
225, 22–30. https://doi.org/10.1016/j.biocon.2018.06.022.
Brotons, L., Thuiller, W., Araújo, M.B., Hirzel, A.H., 2004. Presence-absence versus presence-only modelling methods for predicting bird habitat suitability. Ecography
27, 437–448. https://doi.org/10.1111/j.0906-7590.2004.03764.x.
Calenge, C., 2006. The package “adehabitat” for the R software: a tool for the analysis of space and habitat use by animals. Ecol. Model. 197, 516–519. https://doi.
org/10.1016/j.ecolmodel.2006.03.017.
Carrasco, L., Mashiko, M., Toquenaga, Y., 2014. Application of random forest algorithm for studying habitat selection of colonial herons and egrets in human-
inuenced landscapes. Ecol. Res. 29, 483–491. https://doi.org/10.1007/s11284-014-1147-0.
Carrasco, L., Toquenaga, Y., Mashiko, M., 2017. Balance between site delity and habitat preferences in colony site selection by herons and egrets. J. Avian Biol. 48,
965–975. https://doi.org/10.1111/jav.01255.
Cheng, L., Zhou, L., Wu, L., Feng, G., 2020. Nest site selection and its implications for conservation of the endangered Oriental White Stork Ciconia boyciana in Yellow
River Delta. China Bird. Cons. Int. 30, 323–334. https://doi.org/10.1017/S0959270919000303.
Choudhary, D.N., Ghosh, T.K., Mandal, J.N., Rohitashwa, R., Mandal, S.K., 2013. Observations on the breeding of the Woolly-necked Storks Ciconia episcopus in
Bhagalpur, Bihar, India. Indian Birds 8 (4), 93–94. Https://indianbirds.in/pdfs/IB_8.4_Final.pdf#page=11.
Dhyani, S.K., Kareemulla, K., Ajit, Handa, A.K., 2009. Agroforestry potential and scope for development across agro-climatic zones in India. Indian J. Agrofor. 32,
181–190. http://sa.indiaenvironmentportal.org.in/les/Agroforestry%20potential.pdf.
Dudley, N., Alexander, S., 2017. Agriculture and biodiversity: a review. Biodiversity 18 (18), 45–49. https://doi.org/10.1080/14888386.2017.1351892.
Fandos, G., K´
ery, M., Cano-Alonso, L.S., Carbonell, I., Tellería, J.L., 2021. Dynamic multistate occupancy modeling to evaluate population dynamics under a scenario
of preferential sampling. Ecosphere 12, e03469. https://doi.org/10.1002/ecs2.3469.
Fasola, M., Brangi, A., 2010. Consequences of rice agriculture for waterbird population size and dynamics. Waterbirds 33 (Sp. Issue 1), 160–166. https://doi.org/
10.1675/063.033.s112.
Fasola, M., Ruiz, X., 1996. The value of rice elds as substitutes for natural wetlands for waterbirds in the Mediterranean region. Waterbirds 19 (Sp. pub. 1), 122–128.
https://doi.org/10.2307/1521955.
Fasola, M., Rubolini, D., Merli, E., Boncompagni, E., Bressan, U., 2010. Long-term trends of heron and egret populations in Italy, and the effects of climate, human-
induced mortality, and habitat on population dynamics. Pop. Ecol. 52, 59–72. https://doi.org/10.1007/s10144-009-0165-1.
Fielding, A., Bell, J., 1997. A review of methods for the assessment of prediction errors in conservation presence/absence models. Environment Conservation.
Cambridge University Press, pp. 38–49. https://www.jstor.org/stable/44519240.
Fischer, J., Hartel, T., Kuemmerle, T., 2012. Conservation policy in traditional agricultural landscapes. Cons. Lett. 5, 167–175. https://doi.org/10.1111/j.1755-
263X.2012.00227.x.
Fischer, J., Abson, D.J., Bergsten, A., Collier, N.F., Dorresteijn, I., Haspach, J., Hylander, K., Schultner, J., Senbeta, F., 2017. Reframing the food-biodiversity
challenge. Trends Ecol. Evol. 32, 335–345. https://doi.org/10.1016/j.tree.2017.02.009.
Sundar, K.S.G., 2020. Special Section Editorial: Woolly-necked Stork – a species ignored. SIS Cons. 2, 33–41. https://storkibisspoonbill.org/wp-content/uploads/
2021/04/2020SpecialSectionEdit_Sundar_Final.pdf.
Survey of India, 2007. Topographic Sheets from the Open Map Series, 1:50, rst ed. Survey of India, Hathibarkala estate, Dehradun, India.
Ghimire, P. , Pandey, N. , Timilsina, Y.P. , Bista, B.S. , Sundar, K.S.G. in press. Woolly-necked Stork activity budget in lowland Nepal’s farmlands: the inuence of
wetlands, seasonal crops, and human proximity. Waterbirds.
Greenwell, B. , 2018. Partial dependence plots, Package pdp, version 0.7.0. https://bgreenwell.github.io/pdp/index.html.
Gula, J., 2020. It is time for the Saddle-billed Stork to be on our radar. SIS Cons. 2, 7–9. https://storkibisspoonbill.org/wp-content/uploads/2021/04/2020Opinion_
Gula_Final.pdf.
Gula, J., Sundar, K.S.G., Dean, W.R.J., 2020. Known and potential distributions of the African Ciconia microscelis and Asian C. episcopus Woollyneck Storks. SIS Cons. 2,
80–95. https://storkibisspoonbill.org/wp-content/uploads/2021/04/2020Gula_Final.pdf.N.
Haase, P., 1995. Spatial pattern analysis in ecology based on Ripley’s K-function: introduction and methods of edge correction. J. Veg. Sc. 6, 575–582. https://doi.org/
10.2307/3236356.
Hancock, J.A., Kushlan, J.A., Kahl, M.P., 1992. Storks, Ibises and Spoonbills of the World. Academic Press, Harcourt Brace Jovanovich, Publishers, New York.
Handa, A.K., Chhavi, S., Chavan, S.B., Dhillon, R.S., Ahlawat, K.S., Rizvi, R.S., 2020. Agroforestry in Haryana: status and way forward. Indian J. Agrofor. 22, 1–10.
Hasan, M.T., Ghimire, P., 2020. Conrmed breeding records of Asian Woollyneck Ciconia episcopus in Bangladesh. SIS Conserv. 2, 47–49. https://storkibisspoonbill.
org/wp-content/uploads/2021/04/2020Hasan_Final.pdf.
Higgins, K.F., Naugle, D.E., Forman, K.J., 2002. A case study of changing land use practices in the Northern Great Plains, U.S.A.: an uncertain future for waterbird
conservation. Waterbirds 25 (Sp. Pub. 2), 42–50. https://www.jstor.org/stable/1522450.
Hothorn, T. , Winell, H. , Hornik, K. , van de Wiel, M.A. , Zeileis, A. , 2021. Package ‘coin’. Conditional inference procedures in a permutation test framework. http://
coin.r-forge.r-project.org. (Accessed 22 June 2021).
Hume, A.O., 1889. The Nests and Eggs of Indian Birds, Second ed. R.H. Porter,, London, pp. 268–269.
Jackson, P., 1999. The Delhi Sultanate: A Political and Military History. Oxford University Press, Cambridge, UK.
Katuwal, H.B., Baral, H.S., Sharma, H.P., Quan, R.-C., 2020. Asian Woollynecks are uncommon on the farmlands of lowland Nepal. SIS Conserv. 2, 50–54. https://
storkibisspoonbill.org/wp-content/uploads/2021/04/2020Katuwal_Final.pdf.
Kim, J.H., Park, S., Kim, S.H., Kang, K., Waldman, B., Lee, M.H., Yu, M., Yang, H., Chung, H.Y., Lee, E.J., 2020. Structural implications of traditional agricultural
landscapes of the functional diversity of birds near the Korean Demilitarized Zone. Ecol. Evol. 10, 12973–12982. https://doi.org/10.1002/ece3.6880.
Kittur, S., Sundar, K.S.G., 2020. Density, ock size and habitat preference of Woolly-necked Storks in agricultural landscapes of South Asia. SIS Cons. 2, 71–79.
https://storkibisspoonbill.org/wp-content/uploads/2021/04/2020Kittur_Final.pdf.
Koju, R., Maharjan, B., Gosai, K.R., Kittur, S., Sundar, K.S.G., 2019. Ciconiiformes nesting on trees in cereal-dominated landscapes: importance of scattered trees for
heronries in lowland Nepal. Waterbirds 42, 355–365. https://doi.org/10.1675/063.042.0401.
Kushlan, J.A., 2012. A history of conserving colonial waterbirds in the United States. Waterbirds 35, 608–625. https://doi.org/10.1675/063.035.0410.
Li, D., Chen, S., Lloyd, H., Zhu, S., 2013. The importance of articial habitats to migratory waterbirds with a natural/ articial wetland mosaic, Yellow River Delta,
China. Bird. Cons. Int. 23, 184–198. https://doi.org/10.1017/S0959270913000099.
Lin, H.-S., Cooke, S.J., Wolter, C., Young, N., Bennett, J.R., 2020. On the conservation value of historic canals for aquatic ecosystems. Biol. Conserv. 251, 108764
https://doi.org/10.1016/j.biocon.2020.108764.
L´
opez-Pomares, A., L´
opez-Iborra, G.M., Martín-Cantarino, C., 2015. Irrigation canals in a semi-arid agricultural landscape surrounded by wetlands: their role as a
habitat for birds during the breeding season. J. Arid Environ. 118, 28–36. https://doi.org/10.1016/j.jaridenv.2015.02.021.
Luzuriage-Neira, N., Cuich´
an, C., da Costa, M., Reyes, N., 2021. Observations on the breeding biology of the Andean Ibis (Thereisticus branickii) in Ecuador. Ornithol.
Res. 29, 22–28. https://doi.org/10.1007/s43388-021-00044-3.
Manly, B.F.J., McDonald, L.L., Thomas, D.L., McDonald, T.L., Erickson, W.P., 2004. Resource Selection by Animals. Statistical Design and Analysis for Field Studies,
Second ed. Kluwar Academic Publishers, New York, USA.
Manzoor, S.A., Grifths, G., Lukac, M., 2018. Species distribution model transferability and model grain size – ner may not always be better. Sci. Rep. 8 (2018),
7168. https://www.nature.com/articles/s41598-018-25437-1.
Mehta, K., 2020. Observations of Woolly-necked Stork nesting attempts in Udaipur city, Rajasthan, India. SIS Conserv. 2, 68–70. https://storkibisspoonbill.org/
publications/sis-conservation-issue-2/.
Milliken, G.A., Johnson, D.E., 2009. Analysis of Messy Data, Second ed. CRC Press, Taylor and Francis Group, New York.
Moreira, F., Martins, R.C., Catry, I., D’Amico, M., 2018. Drivers of power line use by white storks: a case study of birds nesting on anthropogenic structures. J. Appl.
Ecol. 55, 2263–2273. https://doi.org/10.1111/1365-2664.13149.
S. Kittur and K.S.G. Sundar
Global Ecology and Conservation 30 (2021) e01793
16
Nachuha, S., Quinn, J.L., 2010. The distribution of colonial waterbirds in relation to a Ugandan rice scheme. Waterbirds 35, 590–598. https://doi.org/10.1675/
063.035.0408.
Ofce of the Registrar General & Census Commissioner, India, 2011. District census handbook. Rohtak and Jhajjar, Census of India,, p. 2011. https://censusindia.gov.
in/2011census/dchb/DCHB_HaryanaA.html.
Pandey, D.N., 2007. Multifunctional agroforestry systems in India. Curr. Sci. 92, 455–463. https://www.jstor.org/stable/24097558.
Parkes, M.L., Mora, M.A., Feagin, R.A., 2012. Using scale, cover type and GIS to evaluate nuisance cattle egret colony site selection. Waterbirds 35, 56–63 https:/doi.
org/10.1675/063.035.0106.
Perfecto, I., Vandermeer, J., 2010. The agroecological matrix as alternative to the land-sparing/ agriculture intensication mode. Proc. Natl. Acad. Sci. 107,
5786–5791. https://doi.org/10.1073/pnas.0905455107.
Prabhakaran, S. , 2016. Performance analysis and companion functions for binary classication models. Package ‘InformationValue’ v. 1.2.3. http://r-statistics.co/
Information-Value-With-R.html.
Purbasha, B., 2017. The Woolly-necked Stork Ciconia episcopus re-using an old Grey Heron Ardea cinerea nest. Indian Birds 13 (6), 165. http://indianbirds.in/pdfs/IB_
13_6_Banerjee_WoollyneckedStork.pdf.
Ricciardi, V., Ramankutty, N., Mehrabi, Z., Jarvis, L., Chookolingo, B., 2018. How much of the world’s food do smallholders produce? Glob. Food Secur. 17, 64–72.
https://doi.org/10.1016/j.gfs.2018.05.002.
Ricciardi, V., Mehrabi, Z., Wittman, H., James, D., Ramankutty, N., 2021. Higher yields and more biodiversity on small farms. Nat. Sustain (2021). https://doi.org/
10.1038/s41893-021-00699-2.
Singh, A., 2011. Estimating long-term regional groundwater recharge for the evaluation of potential solution alternatives to waterlogging and salinisation. J. Hydrol.
406, 245–255. https://doi.org/10.1016/j.jhydrol.2011.07.004.
Singh, H., 2015. Asian Woollyneck Ciconia episcopus breeding in western Rajasthan. India Bird. 24, 130–131.
Space Application Centre (ISRO), 2010. National Wetland Atlas: Haryana, SAC/RESA/AFEG/NWIA/ATLAS/15/2010. Space Applications Centre (ISRO), Ahmedabad,
India, p. 148. https://vedas.sac.gov.in/vedas/downloads/atlas/Wetlands/NWIA_Haryana_Atlas.pdf.
Stanton, R.L., Morrissey, C.A., Clark, R.G., 2018. Analysis of trends and agricultural drivers of farmland bird declines in North America: a review. Agric. Ecosyst.
Environ. 254, 244–254. https://doi.org/10.1016/j.agee.2017.11.028.
Sueltenfuss, J.P., Cooper, D.J., Knight, R.L., Waskom, R.M., 2013. The creation and maintenance of wetland ecosystems from irrigation canal and reservoir seepage in
a semi-arid landscape. Wetlands 33, 799–810. https://doi.org/10.1007/s13157-013-0437-6.
Sundar, K.S.G., 2004. Group size and habitat use by Black-necked Storks Ephippiorhynchus asiaticus in an agriculture-dominated landscape in Uttar Pradesh, India.
Bird. Cons. Int. 14, 323–334 https://doi.org/10/1017/S0959270904000358.
Sundar, K.S.G., 2006. Flock size, density and habitat selection of four large waterbird species in an agricultural landscape in Uttar Pradesh, India. Waterbirds 29,
365–374. https://doi.org/10.1675/1524-4695(2006)29[365:fsdahs]2.0.co;2.
Sundar, K.S.G., 2011. Agricultural intensication, rainfall patterns, and large waterbird breeding success in the extensively cultivated landscape of Uttar Pradesh,
India. Biol. Cons. 144, 3055–3063. https://doi.org/10.1016/j.biocon.2011.09.012.
Sundar, K.S.G., Kittur, S., 2012. Methodological, temporal and spatial factors affecting modelled occupancy of resident birds in the perennially cultivated landscape of
Uttar Pradesh, India. Landsc. Ecol. 27, 59–71. https://doi.org/10.1007/s10980-011-9666-3.
Sundar, K.S.G., Kittur, S., 2013. Can wetlands maintained for human use also help conserve biodiversity? Landscape-scale patterns of bird use of wetlands in an
agricultural landscape in north India. Biol. Cons. 168, 49–56. https://doi.org/10.1016/j.biocon.2013.09.016.
Sundar, K.S.G., Subramanya, S., 2010. Bird use of rice elds in the Indian subcontinent. Waterbirds 33 (Sp. pub. 1), 44–70. https://doi.org/10.1675/063.033.s104.
Sundar, K.S.G., Chauhan, A.S., Kittur, S., Babu, S., 2015. Wetland loss and waterbird use of wetlands in Palwal district, Haryana, India: the role of agriculture,
urbanization and conversion to sh ponds. Wetlands 35, 115–125. https://doi.org/10.1007/s13157-014-0600-8.
Sundar, K.S.G., Koju, R., Maharjan, B., Marcot, B.G., Kittur, S., Gosai, K.R., 2019. First assessment of factors affecting the breeding success of two stork species in
lowland Nepal using Bayesian Network analyses. Wildfowl 69, 45–69. https://wildfowl.wwt.org.uk/index.php/wildfowl/article/view/2690.
Thabethe, V., Downs, C.T., 2018. Citizen science reveals widespread supplementary feeding of African woolly-necked storks in suburban areas of KwaZulu-Natal,
South Africa. Urban Ecosyst. 21, 965–973. https://doi.org/10.1007/s11252-018-0774-6.
Thabethe, V., McPherson, S., Downs, C.T., 2021. Diet of nestling African woolly-necked storks in suburban areas of KwaZulu-Natal, South Africa. Afr. J. Ecol. 59,
769–772. https://doi.org/10.1111/aje.12859.
Thabethe, V. , 2018. Aspects of the ecology of African Woolly-necked Storks (Ciconia microscelis) in an anthropogenic changing landscape in KwaZulu-Natal, South
Africa. Unpublished thesis, School of Life Sciences, College of Agriculture, Science and Engineering, University of KwaZulu-Natal, Pietermaritzburg Campus.
Tourenq, C., Benhamou, S., Sadoul, N., Sandoz, A., Mesl´
eard, F., Martin, J.-L., Hafner, H., 2009. Spatial relationships between tree-nesting heron colonies and rice
elds in the Camarague, France. Auk 121, 192–202. https://doi.org/10.1093/auk/121.1.192.
Treinys, R., L˜
ohmus, A., Stonˇ
cius, D., Skuja, S., Drobelis, E., ˇ
Sableviˇ
cius, B., Rumbutis, S., Dementaviˇ
cius, D., Naruˇ
seviˇ
cius, V., Petraˇ
sca, A., Augutis, D., 2008. At the
border of ecological change: status and nest sites of the Lithuanian Black Stork Ciconia nigra population 2000-2006 versus 1976-1992. J. Ornithol. 149, 75–81.
https://doi.org/10.1007/s10336-007-0220-7.
Tryjanowski, P., Sparks, T.H., Jerzak, L. (Eds.), 2006. The White Stork in Poland. Bogucki Wydaw, Pozna´
n, Poland.
Tscharntke, T., Clough, Y., Wanger, T.C., Jackson, L., Motzke, I., Perfecto, I., Vandermeer, J., Whitbread, A., 2012. Global food security, biodiversity conservation and
the future of agricultural intensication. Biol. Cons. 151, 53–59. https://doi.org/10.1016/j.biocon.2012.01.068.
U.S. Geological Survey, 2018. Landsat 8 Data Products. Geological Survey, U.S. Department of the Interior, Washington D.C., USA. http://glovis.usgs.gov/.
Umrani, R., Jain, C.K., 2010. Agroforestry. Systems and Practices. Oxford University Press, Jaipur, India. ISBN 978-93-80179-17-9.
Vaghela, U., Sawant, D., Bhagwat, V., 2015. Woolly-necked Storks Ciconia episcopus nesting on mobile-towers in Pune. Maharashtra Indian Birds 10 (6), 154–155.
https://indianbirds.in/pdfs/IB_10_6_VaghelaETAL_WoollyneckedStorks.pdf.
Vergara, P., Aguirre, J.I., Fargallo, J.A., D´
avila, J.A., 2006. Nest-site delity and breeding success in White Storks Ciconia ciconia. Ibis 148, 672–677. https://doi.org/
10.1111/j.1474-919X.2006.00565.x.
Vlachos, C.G., Bakaloudis, D.E., Alexandrou, O.G., Bontzorlos, V.A., Papakosta, M.A., 2008. Factors affecting the nest site selection of the black stork, Ciconia nigra in
the Dadia-Lefkimi-Soui National Park, north-eastern Greece. Folia Zool. 57 (3), 251–257.
Win, M.S., Yi, A.M., Myint, T.S., Khine, K., Po, H.S., Non, K.S., Sundar, K.S.G., 2020. Comparing abundance and habitat use of Woolly-necked Storks Ciconia episcopus
inside and outside protected areas in Myanmar. SIS Cons. 2, 96–103. https://storkibisspoonbill.org/wp-content/uploads/2021/04/2020Win_Final.pdf.
Zawardzki, G., Zawadzka, D., Sołtys, A., Drozdowski, S., 2020. Nest-site selection by the white-tailed eagle and black stork – implications for conservation practice.
For. Ecosyst. 7, 59. https://doi.org/10.1186/s40663-020-00271-y.
S. Kittur and K.S.G. Sundar
Fig. A.1. Brood sizes of Woolly-necked Storks in Jhajjar and Rohtak districts, Haryana
observed between 2016 and 2020. Sample sizes are presented in Table 1.
Fig. A.2. Ripley’s K-function showing expected (blue) and observed (red) spatial patterns of
Woolly-necked Stork nest locations in Jhajjar and Rohtak districts, Haryana. Dotted lines are
95% confidence intervals computed from 99 permutations. Patterns are statistically
significant in all graphs (except for 2018) since observed values lie outside of the 95% CI
values.
Fig. A.3. Partial dependence plots following logistic regression to assess habitat suitability of
Jhajjar and Rohtak districts, Haryana, for locating nests by Woolly-necked Storks. Y-axis
measures are distances (in m) to listed variables. For “All”, we used all monitored nests (N =
298).
Fig. A.4. Area under ROC curves for logistic regressions that used all four variables of
interest – distances to irrigation canals, human habitation, tree patches and wetlands –
towards building habitat suitability of Jhajjar and Rohtak districts, Haryana, for nesting
Woolly-necked Storks.
Table A.1. Percentage of areas with different levels of habitat suitability for breeding
Woolly-necked Storks across Jhajjar and Rohtak districts, Haryana. The spatial spread of the
different levels of suitability are provided in Fig. 3. The study area was 3,579 km2.
Year
Suitability
<0.25
0.25-0.5
0.5-0.75
>0.75
2016
12
37
40
11
2017
30
22
28
19
2018
23
18
21
37
2019
22
34
34
10
2020
15
33
37
15
Combined
16
36
43
5
Table A.2. Comparing distances to variables with two kinds of Woolly-necked Stork nests in
Jhajjar and Rohtak districts, Haryana. Boxplots are provided only for variables that had
significant differences in the combined data set.
a. Statistical differences (Prob(P) of permutation tests) of distances from nests with site
fidelity versus nest sites used once. Values are in bold when the test showed statistical
significance (P ≤ 0.05). The accompanying boxplot below the table shows the difference
in distance to wetlands of nests that were used only once (“0”) versus nests where
Woolly-necked Storks showed site fidelity (used > 1year; “1”).
Year
Distance to:
Irrigation
canal
Human
habitation
Tree patches
Wetlands
2016
0.39
0.98
0.69
0.92
2017
0.92
0.4
0.86
0.44
2018
0.84
0.78
0.64
0.75
2019
0.37
0.12
0.59
0.14
2020
0.64
0.99
0.66
0.98
Combined
0.61
0.47
0.9
0.06
b. Statistical differences (Prob(P) of permutation tests) of distances from nests on pylons
versus nests on trees. Values are in bold when the test showed statistical significance (P ≤
0.05). The accompanying boxplot below the table shows the difference in distance to
irrigation canals of nests on pylons versus nests on trees.
Year
Distance to:
Irrigation
canal
Human
habitation
Tree patches
Wetlands
2016
0.13
0.31
0.73
0.13
2017
0.41
0.99
0.37
0.34
2018
0.19
0.24
0.39
0.44
2019
0.14
0.99
0.82
0.88
2020
0.86
0.80
0.011
0.99
Combined
0.043
0.198
0.62
0.56
