Post-monsoon bird assemblages in rural and riverine environments of Northern Howrah, West Bengal, India: A spatio-temporal approach

The primary goal of this study was to evaluate avian assemblages in four different land types (Wetland Associated Bamboo Forest, Agricultural Land, Wetland Associated Agricultural Land and Riverside Agricultural Land) located in a flood prone area of South Bengal during the post monsoon months of 2020. Surveys were conducted for 48 days in a land type and time dependent manner to explore the changes in abundance, dominance, evenness and species richness. A total of 18654 observations belonging to 16 orders, 44 families, 70 genera and 81 species were recorded. Despite having lowest total count and third species rich area, Bamboo Forest emerged as the most diverse area with the highest Shannon index (3.37), Evenness index (0.4616) and lowest dominance index (0.06108). Bamboo forests exhibited the maximum values of Simpson 1-D (0.9389), Brillouin index (3.312), Menhinick Index (1.261), Equitability (0.8134) and alpha diversity (Fisher’s alpha 11.74). Having maximum number of species occurrence (69), Riverside agricultural land was found to be the least diverse land type (Shannon index 2.847) showing highest value of dominance (0.1203). December attracted the most diverse (Shannon index 3.474) bird community and October showed the lowest value of Shannon index (2.919) with maximum value of dominance (0.123). We conclude that protection of natural habitats like bamboo forests, wetland and tree cover in small villages and proper management of agricultural lands is necessary for promoting species diversity and evenness in a rural environment.


Introduction
India is home for 1263 species of birds belonging to 23 orders, 107 families and 498 genera and harbours nearly 12.5% of all birds of the world (Praveen et al., 2016). The avifaunal diversity of West Bengal is very rich with 861 species including some extremely rare ones (Rahmani et al., 2016). Knowledge gathered through the study of bird communities can be used as bioindicators to monitor the health of an environment or ecosystem as well as environmental hazards including several pollution events (Padoa-Schioppa et al., 2006;Amat and Green 2010;Martínez, 2012).
Properly managed agricultural lands are important components for biodiversity conservation and are equally important as the forest patches that harbour a large number of species as well as serve as safe corridors for the dispersal of wildlife between patches (Perfecto and Vandermeer, 2002;Norris, 2008;Rodrigues et al., 2013;Deikumah et al., 2017). Agricultural lands with abundant native tree cover exhibit structural heterogeneity and provide habitat and resources for several species (Deikumah et al., 2017). Elucidating the biodiversity at riverine landscapes often reflects the global conservation importance (Ormerod, 1999;Buckton and Ormerod, 2002). Despite several comprehensive studies on diversity, species distribution and ecology of birds in riverine landscapes, still there is lacking of vivid knowledge on global patterns of species richness in riverine landscapes (Buckton and Ormerod, 1997;Buckton and Ormerod, 2002;Buckton and Ormerod, 2008;Sau et al., 2018;Sinha et al., 2019).
A number of previous studies have demonstrated avifaunal assemblages in different types of agricultural lands and suburban landscape of South Bengal (Hossain and Aditya, 2014;Roy et al., 2016;Mukhopadhyay and Mazumdar, 2017;Sau et al., 2018;Mukhopadhyay and Mazumdar, 2019;Chakraborty et al., 2020;Ghosh et al., 2021). However, very few studies (Sau et al., 2018) have conducted the investigation on avifaunal assemblages in river adjoining agricultural lands in South Bengal and further compared the results with other land types located away from the river. Several studies have reported the avifauna from different urban areas of Howrah district. Majumdar et al. (1992) have reported 540 species belonging to 210 genera, 60 families and 16 orders from West Bengal including 22 species from Howrah district. However, most of these studies were conducted in the Santragachi Jheel (a lake near Santragachi railway station) of Howrah district. Mazumdar et al. (2007) have reported 16 species of waterbirds and Patra et al. (2010) have recorded bird 33 species belonging to 8 families and 23 genera in the Santragachi Jheel of Howrah district. A study by Khan (2010) has described temporal changes to the community structure of 15 species of migratory waterbirds in the lake. Biswas and Banerjee (2016) reported a total of 38 species of birds belonging to 23 families under 12 orders and Palit (2018) has identified 25 species of water birds in the same lake of the district. Recently Chakraborty et al (2020) has described assemblages of 43 bird species belonging to 12 orders and 26 families at two urban areas (Liluah and Belur) of Howrah district. A study from 17 freshwater wetlands of rural and urban areas of Howrah has reported 54 species of wetland dependent or wetland associated birds from 14 families (Nandi et al., 1999). Only two species of water birds (Ardeola grayii and Egretta garzetta) have been described so far from the Udaynarayanpur block of Howrah district (Nandi et al., 1999). To our knowledge, no comprehensive study on avian assemblages has so far been reported from any flood prone rural land type of the Howrah district in South Bengal.
The aim of this study was therefore to explore avian assemblages in four different land types located near riverine landscape in a flood prone block of West Bengal. In this paper, we ask whether diversity of avifauna in the rural environment changes in a land type as well as time dependent manner in the post monsoon season. We further explored changes in abundance, dominance, evenness and species richness under the same conditions.

Study area
The study was conducted in four main land cover types (WABF: Wetland Associated Bamboo Forest, AGL: Agricultural Land, WAAL: Wetland Associated Agricultural Land and RAL: Riverside Agricultural Land) located in four villages (Sultanpur, Sitapur, Baira Kurchi and Jangalpara Belgram) of Udaynarayanpur block of Howrah district (Figures 1 and 2). Udaynarayanpur community development block (22°43'01''N 87°58'30''E) is the northernmost part of Howrah district of West Bengal and belongs to the Uluberia subdivision of the district. The block (124.80 km2) is bounded by the Hooghly district on three sides and is located 43 km from the district headquarters. One of the most flood-prone regions of the district, Udaynarayanpur is located in between two rivers, Damodar and Mundeswari and the present study area is located very close to river Damodar. The study area encompasses a number of swampy areas along with ponds, jheels which are dominated by amphibious macrophytes including some family members of Cyperaceae and Gramineae (Nandi et al., 1999 RAL: These are agricultural lands located on the West River bank of the Damodar river in Jangalpara Belgram village. These plots are dominated by paddy fields interspersed with scrubland, tree cover and water bodies. Some of the important tree species of the area are Areca catechu, Albizia lebbeck, Dalbergia sissoo, Ficus racemose, Borassus flabellifer, Acacia auriculiformis etc. Further, the river terraces, floodplain of Damodar are also part of this habitat type that harbour a number of fish and invertebrates and attract a variety of birds. WAAL: These are wetland associated agricultural lands located in flood prone areas of Baira Kurchi and Sultanpur Village. These lands remain submerged for months after the monsoon season and are dominated by paddy fields with several wetland vegetations like Aeschynomene indica, Cyperus spp, Ipomoea aquatica, Typha elephantina, Marsilea quadrifolia, Trapa bispinosa etc. (Nandi et al., 1999). The study was conducted during the post-monsoon months (October, November and December) between 01 October and 31 December 2020. Vegetation and birds were surveyed using a grid-based method as described by Haney et al. (2008). Equal number (four per land type) of survey grids (250 m × 250 m) were established in all four land cover types with a total of 16 survey grids ( Figure 1; Table 1). Each of these survey grids were subdivided into 25 grid cells each measuring 50m × 50m. A 25 m habitat buffering zone was maintained around each survey grid to reduce the effects of edge. A minimum distance of 500m between two successive survey grids were allocated and the survey of both birds and vegetation data was completed by a single investigator. Each land type (four survey grids) was visited four times a month (once per survey grid) with a total of 12 survey days and all four land types received a total of 48 survey days. Birds were surveyed from dawn to mid-morning during each week of October, November and December using the methods previously described by Haney et al. (2008). All birds detected visually or acoustically were recorded by advancing slowly along the line transect within 25m to the front and on both sides of the line transect. The birds were identified using a binocular and field guides (Ali and Ripley 1987;Grimmett et al., 2015). Bird surveys were avoided during adverse weather conditions such as low cloud, high winds, rainfall fog and strong winds. For vegetation survey (tree with diameter at breast height ≥10 cm), ten of the 25 grid cells were randomly selected placing transects (50 m) through the middle of the grids. Five main variables (percentage) viz Tree cover (TC), Bamboo cover (BC), Shrub (SH), Paddy cover (PC) and Wetland cover (WC) were measured in order to elucidate the relationships between bird assemblages and their environment through Canonical correspondence analysis (CCA). We used three 2 x 2 m quadrats for the estimation of vegetation percentage and average values from the three quadrats were used for CCA. Bird species having fewer than 10 observations among all the counts were eliminated from the present study.

Data analysis
Species with less than 10 counts throughout the study were excluded from all our analyses. In order to investigate the relationship among the bird assemblages and habitat variables a CCA was performed. Environmental variables (percentage) were subjected to Z-score transformation and species abundances to square root transformation prior to CCA to normalize distributions. All box plots, individual-based rarefaction curves CCA and calculation of month and land cover -wise diversity and evenness indices were performed using PAST version 4.03 software (Hammer et al., 2001) and other graphical representations of data were generated using Microsoft Office Excel, 2010. To measure any statistically significant differences between the means of plot and sample-wise group data (species richness and species abundance), one-way analysis of variance (ANOVA) was conducted setting Alpha to 0.05. Post-hoc Tukey HSD (honestly significant difference) between pairs or subgroups after a significant ANOVA result, was performed using Tukey HSD calculator (astatsa.com) to identify the variables that differ significantly. The Venn diagrams showing the number of species among different land types and months were generated using online Venn diagram generator of the Bioinformatics and Systems Biology Group of the Department of Plant Systems Biology, Ghent University (https://bioinformatics.psb.ugent.be/webtools/Venn/).

Results
The present study has recorded a total of 18654 observations of bird occurrence belonging to 16 orders, 44 families, 70 genera and 81 species across different land types during the three months of post-monsoon season ( Figure 3; Table 3). A total of 13 winter migrant (M), 6 local migrant (LM) and 62 resident (RS) species were observed during the entire study (Table 3). Family and Order wise proportional abundance data showed that Passeriformes (58.72%) was the most abundant order and only three orders (Passeriformes, Pelecaniformes and Columbiformes) contributed to nearly 83% of all bird individuals observed ( Figure 3B). Some of the least abundant orders recorded were Anseriformes, Bucerotiformes and Strigiformes. Only two families (Sturnidae and Ardeidae) out of 44 represented nearly 45% of all birds recorded ( Figure 3A). Some of the least abundant families documented were Monarchidae, Turdidae, Caprimulgidae and Paridae. The most abundant species was Gracupica contra followed by Bubulcus ibis, Cypsiurus balasiensis, Motacilla flava, Turdoides striata and others. The 10 most abundant birds of the study constituted more than 67% of the total count.  (Table 2). Our study documented the highest value (3.37) of Shannon index (Shannon_H) in WABF and lowest value (2.847) at RAL (Table 2). Therefore, WABF showed a more evenly (Evenness e^H/S = 0.4616) distributed bird community as compared to the other land types and was the least dominant (Simpson's dominance index D = 0.06108) of all (Table 2). Furthermore, these bamboo forests exhibited the maximum values of Simpson_1-D (0.9389), Brillouin index (3.312), Menhinick's Index (1.261), Equitability_J (0.8134) and alpha diversity (Fisher's alpha 11.74). On the other hand, having maximum number of species occurrence (69), RAL was found to be the least diverse land type (Shannon index 2.847) showing highest value of dominance (0.1203) ( Table 2). Results of the month wise diversity indices revealed that December attracted the most diverse (Shannon index 3.474) bird community in comparison to the other two months ( Table 2). The month of October showed the lowest value of Shannon index (2.919) with maximum value of dominance (0.123) ( Table 2). The number of common species occurrence among four land types and three months were 49 and 67 respectively ( Figure 4A and B). Avian species richness was astonishingly similar between RAL (69) and WAAL (68) and between WABF (63) and AGL (61) ( Figure 4A). However, species richness per plot (survey grid) among these four land types were significantly different from each other ( Figure 5A). Maximum difference in species richness per plot (F=9.28, p<0.01) was recorded between AGL and RAL. Month wise data showed similar species richness among all three months ( Figure 4B). However, species richness per sample showed significant difference (F=4.95, p<0.01) between October and December ( Figure 5C). We also recorded significant differences in plot wise species abundance with maximum difference (F= 9.69, p<0.01) between WABF and RAL ( Figure 5B). However, sample wise species abundance data showed no significant difference among three months during the study ( Figure 5D).
Among these four land types, RAL harboured 10 unique species (Lonchura striata, Plegadis falcinellus, Lonchura malacca, Dendrocygna javanica, Charadrius dubius, Anthus rufulus, Actitis hypoleucos, Ceryle rudis, Tringa glareola, Motacilla citreola) and WABF three (Terpsiphone paradisi, Geokichla citrina, Dendrocopos mahrattensis) which were not found in other two land types ( Figure 4B).  The feeding guild analysis of pooled data from all land types revealed that the study area is dominated by insectivore (38.27%, 31 species) followed by carnivore (25.92%, 21 species), omnivore (18.51%, 15 species) and other species (Figure 6; Table 3). Land type wise data showed that WABF harboured the maximum number of insectivore (38.09%), AGL had the highest proportion of frugivore (8.19%), nectivore (3.27%) and omnivore (21.31%), RAL represented the maximum number of granivore (10.14%), WAAL sheltered the highest proportion carnivore (26.47%) ( Figure 6A). Further a month wise analysis showed that the October represented the maximum percentage of frugivore (6.75%), granivore (9.45%) and nectivore (2.70%), November attracted the maximum percentage of insectivore (40%) and December harboured the maximum number of carnivore (25.64%) and (omnivore (19.23%) ( Figure 6B). In order to assess habitat and month wise distribution and differences in relative species abundance, species richness and species evenness, rank abundance curves were generated using log transformed proportional abundance (percentage) data of each land type and month (Figure 7). Land types in WABF and WAAL exhibited a very similar pattern ( Figure 7A) indicating less differences in bird community structure in these areas. WABF showed maximum evenness in comparison to all other lands. On the other hand, AGL and RAL followed a similar pattern with RAL being the most speciose of all land types ( Figure 7A). Month wise rank abundance curves revealed that October and November shared very similar patterns in species richness and evenness. However, December was the most species rich month having the maximum evenness Figure 7B).   Individual-based rarefaction curves ( Figure 8A) of all birds from four land types indicated that bird assemblages of RAL and WAAL were more diverse than that of WABF and AGL. RAL was the most species rich land type while the least diverse land was AGL. Results from month wise rarefaction curves ( Figure 8B) revealed that species richness in December consistently exceeds the richness of other two months. The results of canonical correlation analysis (Figure 9) revealed that 62.61% of the variance in species abundance was exhibited by only one environmental variable (tree cover) and 37.39% were explained by other four environmental variables. The lowest variance in species abundance was observed in wetland cover. In the present study, four out of five variables were representing vegetation structure that affected the bird assemblage significantly ( Figure 9). The CCA Triplot (Figure 9) revealed that birds like Acridotheres tristis, Caprimulgus macrurus, Dendrocopos mahrattensis, Terpsiphone paradisi and Geokichla citrina tend to co-occur and exhibits positive association with tree and bamboo cover in WABF. However, Ficedula albicilla, Psilopogon haemacephalus, Psittacula krameria, Pelargopsis capensis tend to co-occur near WAAL. Alcedo atthis, Bubulcus ibis, Motacilla alba, Motacilla cinerea showed positive association with paddy cover near AGL. A cluster of ten species (as mentioned earlier) were exclusively associated with river side land and showed no relationship with the variables. Not surprisingly, Hierococcyx varius and Hirundo rustica were observed mostly in the scrub vegetation near wetland of WABF. Figure 9. Canonical Correspondence Analysis (CCA) triplot representing bird species distribution with reference to study plots and environmental variables in Udaynarayanpur, West Bengal, India Vectors indicate direction and strength of relationships between taxa and habitat variables. Environmental variable codes: Tree cover (TC), Bamboo cover (BC), Shrub (SH), Paddy cover (PC) and Wetland cover (WC). Study site/plot codes are given in Table 1 and Species codes are given in Table 3.

Discussion
The present study demonstrates an outline of rural avian assemblages from an extremely flood prone area of South Bengal located in between two rivers. The work provides evidence that areas (e.g. RAL) with highest species richness and maximum total abundance does not always make it to be the most diverse one. On the other hand, less species rich areas (e.g. WABF) in a village might support more bird diversity than others. The findings of the study are important for identification of such small zones and creating awareness for their conservation. This study has also reflected the augmentation of diversity indices in the last month of the post monsoon season and importance of plant composition of an area on the diversity of its avifauna.
In our study the differences in species composition among four land types were more distinct than that of three months of the post monsoon season. This might be explained by the significant habitat heterogeneity (vegetation type and height, wetland abundance, proportion of human habitation etc) between each land types and almost (except last two weeks of December) homogeneous weather patterns (temperature, relative humidity, rainfall, wind etc) during the post monsoon months.
Our results indicated that RAL and WAAL attract more species than the other two land types. This can be attributed to the availability of a heterogenous environment having paddy fields, river bank or submerged land, scrubland, tree cover, water bodies and wetland vegetations. December remained the most species rich month as some of the migratory species arrive late at the start of the Winter and availability of several Winter crops in the fields after the paddy season is over. The effect of human habitation on the total bird abundance has been reflected in the lowest total bird count (2498) of WABF plots in two villages. Other three land types were located far from the human settlements and hence showed significantly higher total bird count. Interestingly total bird counts of December (4978) decreased notably in comparison to the other two months. We presume that a drastic reduction of paddy and water level in the submerged areas in December might have some influences on the lower total bird count.
Our study area is a Doab land, surrounded by two rivers. We documented maximum bird count (~39%) as well as highest species richness near the riverine (RAL) habitats. The principal crop paddy, a variety of vegetables, the riverine canopy and numerous types of invertebrates on riverbed provide higher food availability to attract several birds including some migratory species. The augmented species richness in the riverine landscapes can also be justified by presence of lateral ecotonal linkages as described by previous studies (Ward et al., 1999;Ward and Tockner, 2001;Ward et al., 2002;Buckton and Ormerod, 2002). Furthermore, the physico-chemical complexity and habitat heterogeneity of riverine landscapes might exhibit more species richness than other such habitats with latitudinal trends alone (Arscott et al., 2000;Buckton and Ormerod, 2002). However, being the most species rich and abundant land type for birds, RAL exhibited the lowest diversity indices of all other lands. This has happened due to the lowest species evenness of the area i.e. top ten species occupied more than 70% of total bird abundance. Nearly 1/6th of all species (13) in our study were Winter migrants. Among these four species (Plegadis falcinellus, Actitis hypoleucos, Tringa glareola, Motacilla citreola) were confined to the RAL only and one (Terpsiphone paradisi) to WABF land type. All these migrants arrived in different months, e.g. Motacilla citreola and Terpsiphone paradisi in October, Actitis hypoleucos and Tringa glareola in November and Plegadis falcinellus in December.
Bamboo forests can serve as excellent biodiversity zones and provide food and shelter to many species of mammals, birds, soil organisms, insects and plants making an ideal ecosystem (Lou and Henley, 2010). Previous studies have shown that bamboo habitats can harbour unique avian assemblages including some rare and specialist species in different parts of the world (Guilherme and Santos, 2009;Rother et al., 2013;Socolar et al., 2013;Tu et al., 2020). Despite having lowest total count and third species rich area, WABF emerged as the most diverse area with the highest Shannon index, Evenness index and lowest dominance index. Our results are consistent with available results from of a recent study (Tu et al., 2020) in East Asia which showed that bamboo forest significantly decreases bird numbers as well as species richness but significantly increases the bird species evenness. We found the maximum number of dimensions of environmental heterogeneity in WABF including bamboo forest, wetland, shrubs, several ponds, nearby paddy fields, human habitation, private gardens with a variety of medicinal, ornamental species and fruit crops and a huge number of canopy tree species which might have contributed to enrichment in bird diversity of the land type. Furthermore, the plant diversity of the WABF plots (results not shown) was significantly greater in comparison to the other three land types and thus supported highest bird diversity. One of our WABF plots encompasses a public cemetery of nearly one hectare land, a burial ground of dead people belonging to a particular religion. We have recorded a number of rare species of birds in this undisturbed place during the study. Cemeteries are often regarded as biodiversity islands especially in urban areas. Previous studies have reported a rich avifaunal diversity in these urban burial grounds (Kowarik et al., 2016;Tryjanowski et al., 2017;Csanády and Mošanský 2017;Villaseñor and Escobar, 2019). However, plant communities other than bamboo are facing the threats of massive deforestation. WABF plots in Sultanpur village has recently lost a large number of Areca catechu, Cocos nucifera, Neolamarckia cadamba and Swietenia mahagoni trees which are the most dominant trees in the area that support the survival of many canopy dwelling species. Our study revealed that these wet land associated bamboo forests with other plant composition are of greater importance than other land types for maintaining high bird diversity.
Almost every year the study area witnessed devastating floods driven by the simultaneous release of water from several upstream dams on Damodar river causing submergence of the land area for a long time. The consequences of flooding on aquatic birds can be both beneficial and detrimental. After the flood is over, the food sources become abundant in the area which in turn attract several aquatic bird species to breed (Poiani, 2006). Despite flood causes rapid depletion in the soil macroinvertebrate prey for several aquatic birds, it may help the birds in their prey availability by keeping the upper soil moist and soft to probe for the prey (Ausden et al., 2001). Furthermore, floods may increase the landscape connectivity for aquatic birds and thereby facilitating their dispersal and survival without being trapped in isolated water-bodies (Poiani, 2006). We found disappearance of several species like Anthus rufulus, Gracupica contra, Sturnia malabarica, Nycticorax nycticorax, Ardeola grayii, Bubulcus ibis, Egretta garzetta, Ixobrychus sinensis, Lonchura malacca, Lonchura punctulata, Lonchura striata, Vanellus indicus, Amaurornis phoenicurus, Charadrius dubius, Metopidius indicus, Actitis hypoleucos and many others during the flood time in August and September. However, these birds again reappear in the early weeks of post monsoon session.
We documented a large number of species in the paddy fields. Except one (WABF) most of the land types in our study were dominated by wet paddy cultivation. West Bengal contributes the highest rice production (~13.9%) in the country and South Bengal yields bulk of the total production. A study by Sundar and Subramanya (2010) has shown that nearly 351 bird species occur in the rice fields of the Indian subcontinent which contributes 27% of all species including 22 endemic species of the region.
One of the remarkable findings of our study was complete absence of house sparrow (Passer domesticus), once a very common bird of rural Bengal from all 16 survey grids of the rural study area. Previous studies have reported a steady decline in the house sparrow population in many urban areas in India including West Bengal (Daniels, 2008;Rajashekar and Venkatesha, 2008;Dandapat et al., 2010;Ghosh et al., 2010). However, our findings are in contrast with few previous studies that have reported occurrence of more sparrow population in smaller towns and rural areas in India (Balaji et al., 2013;Roshnath et al., 2018).

Conclusions
The present study provides information about the relationship between species diversity and environmental variables in different land types in the flood prone Udaynarayanpur block of South Bengal. We found that river adjoining agricultural lands (RAL) provide much heterogeneity and hence exhibit more species richness, bird abundance but less diversity and evenness than other land types. A much better diversity and evenness have been recorded in AGL and WAAL land types. However, WABF being the lowest bird abundant site revealed the highest bird diversity and evenness. We conclude that protection of natural habitats like bamboo forests, wetland and tree cover in small villages and proper management of agricultural lands is necessary for promoting species diversity and evenness in a rural environment.