Factors affecting the breeding ecology of the globally threatened Lesser Adjutant ( Leptoptilos javanicus ) in agricultural landscapes of Nepal

. Many threatened birds use the mosaic of agricultural landscapes for foraging and breeding. Despite the reliance of many species on these habitats, few studies have investigated factors influencing the breeding ecology of storks in agricultural landscapes. We assessed site-level variables (tree height and location of nest tree; human habitation or non-human habitation), colony-level variables (colony size and chicks per nest), and landscape-level variables (area of human habitation, wetland area, and distance to the nearest wetland) to understand the factors influencing the breeding ecology of the globally threatened Lesser Adjutant ( Leptoptilos javanicus ) across multiple locations in the agricultural landscape of lowland Nepal during 2019–2020. We monitored 65 active colonies that had 206 active nests in five study sites. Two hundred eighty chicks fledged from these colonies, with 13% (n = 41) chick mortality. Most colonies were in agricultural land (51%) and human habitation (28%). Lesser Adjutant colonies located on tall trees such as Bombax ceiba (57%), Haldina cordifolia (11%), and Ficus religiosa (11%); however, these tree species were used much more than their availability on the landscape. Tree height had a significant positive influence on colony site selection and colony size, whereas colony size positively influenced fledgling success. Measured landscape variables did not have significant relationships with breeding success metrics. The agricultural landscapes of lowland Nepal provided important breeding habitat for Lesser Adjutants, and the suitability of sites with colonies related more to site-level and colony-level than landscape-level variables. Increasing urban development of agricultural landscapes is likely the greatest threat to breeding Lesser Adjutants, with the decline of suitable nesting trees being a potential additional threat. Lowland Nepal's agricultural landscapes support significant breeding populations of Lesser Adjutants that had considerable breeding success, underscoring the urgent need to support traditional agriculture that favors large waterbirds.

ABSTRACT. Many threatened birds use the mosaic of agricultural landscapes for foraging and breeding. Despite the reliance of many species on these habitats, few studies have investigated factors influencing the breeding ecology of storks in agricultural landscapes. We assessed sitelevel variables (tree height and location of nest tree; human habitation or non-human habitation), colony-level variables (colony size and chicks per nest), and landscape-level variables (area of human habitation, wetland area, and distance to the nearest wetland) to understand the factors influencing the breeding ecology of the globally threatened Lesser Adjutant (Leptoptilos javanicus) across multiple locations in the agricultural landscape of lowland Nepal during 2019-2020. We monitored 65 active colonies that had 206 active nests in five study sites. Two hundred eighty chicks fledged from these colonies, with 13% (n = 41) chick mortality. Most colonies were in agricultural land (51%) and human habitation (28%). Lesser Adjutant colonies located on tall trees such as Bombax ceiba (57%), Haldina cordifolia (11%), and Ficus religiosa (11%); however, these tree species were used much more than their availability on the landscape. Tree height had a significant positive influence on colony site selection and colony size, whereas colony size positively influenced fledgling success. Measured landscape variables did not have significant relationships with breeding success metrics. The agricultural landscapes of lowland Nepal provided important breeding habitat for Lesser Adjutants, and the suitability of sites with colonies related more to site-level and colony-level than landscape-level variables. Increasing urban development of agricultural landscapes is likely the greatest threat to breeding Lesser Adjutants, with the decline of suitable nesting trees being a potential additional threat. Lowland Nepal's agricultural landscapes support significant breeding populations of Lesser Adjutants that had considerable breeding success, underscoring the urgent need to support traditional agriculture that favors large waterbirds.

INTRODUCTION
Many bird species are declining more rapidly in agricultural landscapes than other land use types (Stanton et al. 2018, PECBMS 2020. The major reasons for these declines are urbanization, hunting and trade, chemical poisoning, electrocution, agricultural intensification, lack of suitable nesting trees, and varying rainfall patterns or temperature anomalies associated with global climate change (Inskipp and Baral 2010, Mitra et al. 2011, Pearce-Higgins et al. 2015, BirdLife International 2022, Katuwal et al. 2021. Significant declines have been noticed in the global north, where farmers practice large-scale monoculture agriculture with extensive mechanization and chemical use (Stanton et al. 2018, Reif and Vermouzek 2019). There has been relatively little research on species diversity and changes in their abundance in agricultural areas in developing countries. This paucity of research has prevented an understanding of the impacts of agriculture on bird populations in these areas, and there is little detailed information on the factors that influence the suitability of specific landscapes for individual bird species (King et al. 2010).
Agricultural landscapes in developing countries are often heterogeneous because of smaller land holdings (< 2 ha per farmer; Lowder et al. 2016, Ricciardi et al. 2021. The diversity of landscape conditions is due to farmers retaining patches of natural habitats such as wetlands, shrublands, grasslands, and a variety of tree species as patches or scattered single trees , Kittur and Sundar 2021. In South Asia, traditional agricultural practices often include traditional agroforestry (planting and retaining trees) and preservation of common lands (e.g., wetlands; Inskipp and Baral 2010, Sundar and Kittur 2012, Amatya et al. 2018, Kittur and Sundar 2021. Remnant habitat patches in these landscapes has improved landscape-level occupancy of a diverse assemblage of resident birds at landscapes scales (Sundar and Kittur 2012) and has facilitated relatively high breeding success of several large waterbird species (Sundar 2011, Kittur andSundar 2021).
The growing number of studies in cereal-dominated, multicropped landscapes has focused on large waterbirds. For some waterbirds, such as Sarus Cranes (Grus antigone), retaining wetlands in croplands with three to five crops per year improved breeding success relative to crane pairs whose territories had only rice paddies (Sundar 2009). Conversion of croplands and wetlands to urban areas or human habitations permanently removed breeding pairs, reducing the breeding population (Sundar 2011). Black-necked Storks (Ephippiorhynchus asiaticus) also benefitted from the presence of wetlands but tolerated some conversion to urban areas (though they were strongly affected by changing rainfall patterns; Sundar 2011). In contrast, Asian Woolly-necks (Ciconia episcopus) situated nests close to irrigation canals in scattered trees planted far from human habitation and wetlands (Kittur and Sundar 2021). These few available studies have underscored the importance of persisting natural features and artificial structures such as canals introduced to benefit farmers in sustaining impressive populations of breeding large waterbirds on South Asian agricultural landscapes.
The Lesser Adjutant (Leptoptilos javanicus) is another large waterbird species (Fig. 1) that was, until recently, assumed to require undisturbed forests and wetlands for its breeding (Poudyal and Nepal 2010, Inskipp et al. 2016, BirdLife International 2021, Bhattarai et al. 2021. Relatively few studies have been conducted on the ecology of this species (Table 1). The majority of these studies have been anecdotal observations, with a small but growing number of studies led by a priori hypotheses (Table 1). Lesser Adjutants in Nepal represent a small portion of the species' breeding range, yet most research on the species has been conducted in Nepal (Table 1). Lesser Adjutants are usually distributed and breed in lowland Nepal (< 300 m elevation; Inskipp et al. 2016). The largest known breeding population has been discovered in the agricultural matrix of central lowland Nepal (Sundar et al. 2016). Most stork colonies were on trees planted for multifunctional agroforestry, and storks selected the largest trees available on the landscape for nesting . Breeding success was influenced by the interaction of various factors, such as colony size, wetland area, human habitation, and changing crops ). This suggests that variables at multiple spatial scales work together on Lesser Adjutant colonies located in agricultural landscapes to affect overall breeding success. However, it is not clear if findings from one location are transferable to the range of conditions that this species breeds in across lowland Nepal's landscape. Do Lesser Adjutants favor larger forest tree species in areas with more forests? Are colony locations and breeding success related to sitelevel, colony-level, and landscape-level variables similarly across lowland Nepal? Understanding the answers to these questions is critical because this species is globally and nationally Vulnerable (Inskipp et al. 2016, BirdLife International 2021. In this study, we investigated the factors affecting the location of colonies and breeding success of Lesser Adjutants in multiple locations across lowland Nepal, where smallholder agriculture is dominant (Fig. 2). We used several variables measured at site, colony, and landscape scales, including some that have been measured in past studies (Table 1). We then analyzed the impact of these variables to understand whether Lesser Adjutant breeding was (1) influenced more by site-level, colony-level, or Table 1. Summary of studies conducted on Lesser Adjutants (Leptoptilos javanicus). We summarized the studies conducted on nesting and breeding ecology into the following three groups: (1) studies with basic natural history findings such as first record or identifying nest sites or habitat use or short distribution review within the country, (2) studies providing some metrics of breeding variables such as nest/colony size, number of chicks, and fledgling success, and (3) studies with hypothesis testing that explained factors affecting the nesting site or breeding success. landscape-level variables, and (2) whether these variables influenced colony site selection and breeding success similarly in multiple locations.

Study area
We conducted this study in multiple locations across southern lowland parts (< 300 m elevation) of Nepal (Fig. 2). This landscape primarily consists of floodplains of different river systems and is largely suitable for agriculture. Multiple crops are grown in each field over the year, with flooded rice paddies dominating during the monsoon or rainy season (June-September), and wheat, maize, sugarcane, and lentils during the winter (November-February). The fields are kept largely fallow during the hot summers (March-June; Koju et al. 2019, Katuwal et al. 2020. Lowland Nepal's climate is a mix of tropical and subtropical types. The monthly average rainfall for 2015-2019 in the study areas ranges from 80-155 mm and the minimum and maximum temperature for the period was from 18-33 °C, respectively (DHM 2021). Thick fog for a few weeks as well as cold and rain for a few days in December and January are characteristic of the winter season.
The landscape of lowland Nepal is primarily agricultural. However, traditional farming methods are practiced that are characterized by relatively low mechanization, trees retained amid crop fields and along the roads and canals, and retention of community-use wetlands of various sizes on the landscape . Wetlands are retained primarily for use by humans and livestock, and they experience year-long use. Trees are retained for multifunctional agroforestry with farmers growing some species for silviculture (e.g., Bombax ceiba), retaining other species for spiritual reasons (e.g., Ficus religiosa), and maintaining commercially valuable fruit-bearing species as both individual scattered trees and as groves (e.g., Mangifera indica; Koju et al. 2019). Human habitation is restricted to relatively high-density villages, towns, and cities; however, scattered houses or small brick factories have been increasing in agricultural lands in recent years (Katuwal et al. 2022), and human presence occurs in the crop

Fig. 2. Surveyed locations for Lesser Adjutant Stork (Leptoptilos javanicus) breeding colonies in lowland Nepal (country map) and
details of locations where colonies were discovered. Detailed maps show land use, location of colonies, and locations of points that were used to measure available tree species on each landscape. We did not find colonies in Kailali and Chitwan districts (shaded areas).
fields throughout the year . Fields are both rain fed and irrigated by a network of canals that has been increasing in recent years. Irrigation canals lead to the formation a very large number of small wetlands whose hydrology is linked to flooding patterns of the canals (H. B. Katuwal and K. S. Gopi Sundar, personal observations). Forest patches are largely preserved as community forests or protected reserves, and the extent of forests in lowland districts varies greatly (Rimal et al. 2018a, MFE 2019. Landholdings are relatively small (< 1 ha; Katuwal et al. 2021), but farmers retain many very small patches of natural grasslands, degraded scrub, bamboo, and open uncultivated areas interspersed with croplands. Therefore, the overall landscape is diverse and patterned with a relatively high and densely populated human presence throughout the year, but it retains natural features alongside croplands and artificial features, such as canals built to support farming (Katuwal et al. 2022).

Research design
We focused on understanding the breeding ecology of the Lesser Adjutant across lowland Nepal in areas that had different levels of agriculture, forests, wetlands, and human population. We initially included seven different regions for surveys (Fig. 2). Searches for stork colonies, however, located none in Chitwan (excluding core areas of Chitwan National Park) and Kailali districts. We retained the other five regions (Kapilvastu-Rupandehi, Bara, Sarlahi, Udayapur and Morang-Jhapa) for this study (Fig. 2).

Identifying and monitoring nesting colonies
Based on existing information and knowledge gleaned from local experts in each location, we selected parts of the focal districts for detailed surveys to locate Lesser Adjutant colonies. Two authors (H. B. Katuwal and K. S. Gopi Sundar) trained research associates at each site to locate and monitor colonies, observe fates of individual nests, and obtain measurements of trees (both nest trees and those measured to assess availability).
The nest initiation period of Lesser Adjutants in lowland Nepal is from August to October, with chicks remaining in colonies until they fledge in February (Sundar et al. 2016). Therefore, each focal study area was surveyed starting from 15 July 2019, following methods previously used in Nepal to study Lesser Adjutants' breeding biology (Sundar et al. 2016. Motorable road routes were traversed using motorbikes, farmers and local researchers were interviewed, and locations mentioned in previous publications were visited to locate stork colonies. A colony was defined as having at least one chick in a nest or set of nests on trees whose canopies were contiguous, following previous studies (Sundar et al. 2016). At each colony, we recorded its location using a handheld Garmin Global Positioning System (GPSMAP 64s), identified the nest tree species, and measured the nest tree height in meters using a range finder (MILESSEY S9) and diameter at breast height (DBH) in centimeters.
Each colony was visited multiple times (1-2 times per month) until January 2020, when the last remaining chicks fledged. We observed nesting materials being collected from 16 July. We noted the first chick on 18 September (the chick might have hatched before but was not visible from the ground) and noticed the first chick fledging on 27 November. Colony size was the maximum number of nests observed at each colony. Lesser Adjutant pairs were not observed using nests of another pair, as in other stork species such as the White Storks (Ciconia ciconia; Vergara et al. 2006). This made it easy to estimate the sizes of all colonies definitively. At each colony, we recorded the number of chicks that fledged from each nest. Lesser Adjutants' pre-fledged chicks were not observed moving around the colony or using nests of multiple pairs as had been observed in other stork species such as Painted Storks (Mycteria leucocephala; K.S. Gopi Sundar, personal observation). Therefore, counts of surviving and fledged chicks per nest were relatively straightforward. We did not climb trees to observe nests; therefore, we are unable to provide information on clutch size, hatching success (proportion of eggs that successfully hatched), and total breeding success (proportion of eggs and hatched chicks that successfully fledged). In addition to systematically monitored colonies, we recorded 13 additional colonies in other districts on an ad hoc basis until March 2020 (Appendix 1, Fig. A1.1) but did not include those for final analyses because the data sets were not comparable.

Measuring tree availability
We generated random points (for four locations) and systematically located points (Kapilvastu-Rupandehi area) using ArcGIS 10.4 to understand the tree availability for the stork to nest in the study area (Fig. 2). All accessible points were visited and the nearest trees' species name, height, and DBH were recorded. We measured a total of 298 trees of 38 species to determine tree availability.

Mapping land use
We acquired terrain-corrected 30 m resolution satellite images from Landsat-8 OLI (10 November to 5 December 2019) to determine land use from the United States Geological Survey (https:// earthexplorer.usgs.gov) for each focal study area within three sections, Kapilvast-Chitwan, Bara-Sarlahi, and Udayapur-Jhapa.
The geometric accuracy of all collected images was verified. This process was conducted one image at a time, registered, and projected in Universal Transverse Mercator zone 45N. We used the ENVI v5.3 environment for processing all the images (https://www. l3harrisgeospatial.com/). We converted the digital number of images into radiance and applied the flash line-of-sight atmospheric analysis of spectral hypercubes (FLAASH) atmospheric correction model (Rimal et al. 2018b). We extracted land-use and land-cover variables using a support vector machine (Kavzoglu andColkesen 2009, Schneider 2012) and radial basis function with assigned maximum penalty parameters to value 100 following Rimal et al. (2019). We classified land use in images into eight classes, as follows: urban (all built-up areas), agriculture land, forest (all forest types), shrubland, barren land (cliffs, small landslide, bare rocks, and unused lands), sand, water bodies, and grasslands. Overall accuracy was 91% for Kapilvastu-Chitwan, 87% for Bara-Sarlahi, and 90% for Udayapur-Jhapa sections.
Around each colony, using classified satellite imageries, we measured the amount of farmlands and wetlands (in ha) using a buffer of 5 km radius, assuming that storks would use these areas for foraging and that measurements at this scale could influence both colony location or site and breeding success (see . We also measured human habitation (built-up areas in ha) around each colony using a buffer of 1 km radius to assess whether human presence influenced both colony location and breeding success. We also calculated the nearest distance between colonies' locations with land uses (patches > 0.25 ha) such as human habitation, wetland, and farmland. We used these three land-use classes because they have been seen to influence stork breeding (Tourenq et al. 2004, Onmuş et al. 2012).

Data analysis
We assessed tree species selection by Lesser Adjutants with the useavailability framework of Manley et al. (2004), whereby selection metrics were contrasted against Bonferroni's 95% confidence intervals to assess whether a particular species of tree was selected (used in higher proportion relative to availability), avoided (used less relative to availability), or used in similar proportion to availability. We used the WidesI function in the adehabitat package for estimations (Calenge 2006). Using the Wilcoxon rank-sum test (W) and ridgeline density plots in the ggridges package, we compared tree size (height and DBH) of those with colonies and those measured at systematic or random locations to determine whether Lesser Adjutants selected trees of specific sizes to locate colonies (Wilke 2021).
We calculated the mean (± SD) of several colony-level breeding variables including colony size, total number of chicks, chicks per nest, and fledgling success per colony. We estimated chicks per nest by dividing the total number of chicks by the number of nests or colony size in each colony. We considered fledgling success, i.e., the number of chicks fledged per colony, as the proxy for breeding success because it incorporates egg and chick mortality (see Sundar et al. 2019). We confirmed chick mortality by direct observations where possible and asked local people living near the colonies when we could not confirm mortality incidents directly. We performed pairwise Pearson correlation tests of the selected landuse variables and tree height and DBH, and subsequently removed redundant variables. For analysis, we retained tree height, wetland area and human habitations area around colonies, and distance to the wetland from the colony.
We were interested to understand the relative importance of sitelevel variables measured at the scale of the colony (tree height, location of nest tree; human habitation or non-human habitation) and at the scale of the larger landscape around the colony (distance to nearest wetland, area of human habitation, and wetland areas). We were interested in assessing the relative importance of these variables in influencing three discrete aspects of breeding, as follows: (1) nest/colony site selection, (2) colony size, and (3) fledgling success. These three aspects parsed out disparate and critical components of the breeding biology of Lesser Adjutants and, collectively, would allow identification of variables important to each stage of breeding (see also Sundar et al. 2016).
We checked for spatial autocorrelations using Moran's I test in the ape package (Paradis and Schliep 2019) and found no significant spatial autocorrelation for all dependent variables (P > 0.05). We used a generalized linear mixed model (GLMM) to understand which variables influenced colony site selection and colony size. We further added colony-level variables (e.g., colony size and chicks per nest) to identify variables influencing fledgling success. To understand colony site selection, we used 65 random points (equalling the number of colonies) as colony/nest absence from 298 random surveys done to measure tree availability. We used tree height and extracted landscape variables of these points. Therefore, our response variables were, (1) colony site selection (presence/absence), (2) colony size (count), and (3) fledgling success (count). Similarly, our predictors were, site-level (tree height and location of nest tree; human habitation or non-human habitation), colony-level (colony size and chicks per nest), and landscape-level (area of human habitation and wetland area, and distance to the nearest wetland).
In our GLMM analysis, we assigned family binomial distribution for colony site selection and Poisson distribution for colony size and fledgling success. We included study site as the random effect in all models. We checked for overdispersion of our three models using overdisp_fun and found no overdispersion. Then, we used an information-theoretic approach (Burnham and Anderson 2002) and reported model averaging with delta corrected Akaike Information Criteria (ΔAICc) < 4 using the package MuMIn (Barton 2020). We calculated standardized coefficients of each model and considered the importance of each variable as high or significant when its 95% confidence interval (CI) did not overlap with zero (Di Stefano 2004). Because two of our study sites (Kapilvastu-Rupandehi and Morang-Jhapa) had colonies dispersed (≥ 20 colonies in each site) more than the other three areas, we reran the analysis to assess whether factors affecting breeding ecology would be consistent with these two sites. All analysis was performed in R software (R Core Team 2019).
Trees used for nesting by storks varied significantly from available trees in both height (W = 5225, P < 0.001) and DBH (W = 2384.5, P < 0.001). The mean (± SD) height of nest trees (17.07 m ± 5.66) was ~ 5 m more than available trees (12.76 m ± 8.39), whereas DBH of nest trees (98 cm ± 37.78) was twice that of available trees (45.61 cm ± 27.72). These differences were visible at each study site (Fig. 3a and 3b). In addition, the height and DBH of the nesting tree with most colonies (Bombax ceiba) also varied within the study area ( Fig. 3c and 3d).

Table 2. Tree species used by breeding Lesser Adjutants (Leptoptilos javanicus) in different locations across lowland
Nepal. Signs indicate trees that were used significantly more relative to availability (+), used in proportion to availability (0), and avoided or used significantly less relative to availability (-). Trees that were not found in a particular locality do not have a sign. "Other" refers to the combination of all the tree species in a locality that did not have any stork colonies. Numbers of colonies monitored are indicated below location names.
Significance was taken at P ≤ 0.05 level.

Factors influencing breeding ecology
The effects of site-level and colony-level variables were more influential than landscape-level variables on the breeding of the Lesser Adjutant in Nepal (Fig. 4). Tree height had significant positive (95% CI did not overlap with zero) impacts on the colony site selection (Fig. 4a). Location of nest tree (non-human habitation negatively and human habitation positively) also had significant impacts on the colony site selection (Fig. 4a). Tree height also had a significant positive (95% CI did not overlap with zero) influence on the colony size (Fig. 4b). However, colony size had a higher significant positive (95% CI did not with overlap zero) influence on fledgling success relative to other variables (Fig.  4c). The landscape variables did not significantly impact breeding metrics (95% CI overlapped with zero). Notwithstanding statistical significance, the area of human habitations had a slightly negative relationship, whereas the distance to wetland and wetland area had positive relationships to breeding metrics (Fig.  4). For study sites analyzed separately, results were identical with colony site selection and fledgling success. However, colony size had a significant negative relationship with the area of human habitation, whereas tree height had significant positive impacts (95% CI did not overlap with zero; Appendix 1, Fig. A1.2).

DISCUSSION
This is the first study of Lesser Adjutant breeding ecology, where colonies at multiple sites across lowland Nepal were studied simultaneously. Our study provides insights into the importance of nest tree species and height-and colony-level variables on the breeding of the Lesser Adjutant. Our study further demonstrates the value of Nepal's agricultural landscapes for breeding Lesser Adjutants that were incorrectly assumed to require forests and protected wetlands (Poudyal and Nepal 2010, Inskipp et al. 2016, Bhattarai et al. 2021. We were able to confirm that the largest breeding populations are in central Nepal, where previous studies were conducted (see Sundar et al. 2019, Koju et al. 2020. Furthermore, we discovered that other sites in lowland Nepal also have breeding Lesser Adjutants, and the new sites collectively add up to significant breeding populations. In our study, the drivers of mortality (13%) varied by study sites and included, for example, extreme weather events (continuous rain and cold for weeks or wind) in Udayapur (S. Chaudhary, personal communication), tree felling in Kapilvastu-Rupandehi, and hunting in Morang-Jhapa and eastern Nepal, in addition to a variety of unknown causes. In other studies, Large-billed Crows (Corvus macrorhynchos) were identified as a threat (south-east Asia; Clements et al. 2013), though nest failures did not occur during other studies (see Sundar et al. 2016). Hunting of the species has been recorded in east Nepal (S. Thapa and A. Sah, personal communication); however, the impact of tree felling is rarely documented. Clearly, inter-annual and inter-site variations of chick mortalities exist for Lesser Adjutants in Nepal. The fledgling success reported in our study is slightly lower than Sundar et al. (2016) reported at Kapilvastu-Rupandehi, which was also the site that provided the largest sample size for our study. However, fledgling success in our study is higher than that reported by Karki and Thapa (2013) in eastern Nepal, which had a much smaller sample size. These variations are likely driven by sample sizes, although inter-annual variations in breeding success cannot be ruled out.
We did not find colonies in agricultural lands along the west of Kapilvastu-Rupandehi, showing that Lesser Adjutants may have a lower population in the region and may predominantly breed in protected forested parks (Sharma 2006;D. Joshi, personal communication). The total population of Lesser Adjutants in this study was 890, which, along with the scale of the study, is the highest count of the species in Nepal. Our surveys did not cover the entirety of the known breeding distribution range in Nepal, suggesting that the actual breeding population is higher. Because of site-level variations in breeding populations, we are unable to provide a robust extrapolation but suspect that the population is much higher than the < 1000 individuals guesstimated for Nepal (Inskipp et al. 2016).
In our study, tree species and size drove the nest site selection of Lesser Adjutant. Storks had most colonies on Bombax ceiba at all sites, although Haldina cordifolia and Ficus religiosa were also widely used. Retaining these tree species on the landscape for longer periods is essential to provide nesting sites for breeding Lesser Adjutants. A large number of available trees of these species were much smaller than nest trees, suggesting that agroforestry practices that favor larger trees will be beneficial for Lesser Adjutant Storks in the region. Other threats to nest trees were development projects, thunderstorms (two nesting trees fell in our study because of storms), and, very rarely, people removing nest trees in response to the nuisance of nesting birds (G. Sah, personal communication). Lesser Adjutants used native tree species almost entirely and completely avoided the much commonly available exotic Eucalyptus sp. Asian Woolly-necks in India used both native trees and the exotic Eucalyptus sp., suggesting that each stork species has nuanced and idiosyncratic preferences for nest trees (Kittur and Sunder 2021). Taller trees were important for locating colonies, as observed in other stork studies (Hilaluddin et al. 2003. Such a choice likely reduces human-related disturbance and storm-related tree uprooting. Lesser Adjutants made more choices among nest tree characteristics and fewer choices among location of the trees, which were frequently in human settlements. Although colony-level variables, such as colony size, have been documented to influence breeding success (Serrano et al. 2005) and provisioning behavior (Sundar et al. 2016), few studies only include these variables when identifying the factors affecting large waterbird breeding success. Our studies of Lesser Adjutants in Nepal underscore the need to include colonylevel variables. This may be especially true for species that do not have very large colonies because colony-level variables likely influence breeding behaviour. (Minias et al. 2020).
Lesser Adjutants were not influenced greatly by landscape-level variables relative to site-level and colony-level variables. This finding is different from other studies that have shown similar landscape-level variables to have either positive or negative influences on breeding success of waterbirds (Sundar 2011, Janiszewski et al. 2014, Kittur and Sundar 2021. We did note a slight negative influence of area of human habitations on both colony size and fledgling success. However, some colonies were observed within human habitations, possibly because of lack of availability of large trees in agricultural habitats. Kittur and Sundar (2021) showed single-nesting Asian Woolly-necks avoiding human habitations for nesting. Overall, it is possible that the Lesser Adjutants, which have relatively small colony sizes, will be negatively impacted by increased urbanization in Nepal.
Contrary to our predictions, wetland areas around colonies and distance to wetlands weakly influenced breeding metrics. It is not immediately clear why storks were not associated more strongly with wetlands on lowland Nepal. Previous studies have underscored the strong relationship between foraging activity and changing crops, suggesting that crop fields are important foraging habitats for Lesser Adjutants and that the dominant crop fields may exceed the positive influence of wetlands alone (Sundar et al. 2016). It is also possible that the current wetland extent and the number of wetlands on Nepal's agricultural landscapes are high enough to not strongly influence stork breeding. Wetland's influence on the Lesser Adjutants may increase in strength if wetland deterioration continues but is an aspect that requires longer term monitoring.