Impact of water level on spring bird assemblages in a natural river valley in central Europe

Abstract Water level appears to be crucial for waterbirds, especially in the changeable conditions prevailing in river valleys. One of the best preserved Central European rivers is the Biebrza in north-eastern Poland, which seasonally overflows in spring, creating optimal habitats for various waterbirds. We examined a number of environmental factors which could strongly influence different foraging groups and bird assemblages with respect to water surface area and depth. Number of waterbirds peaked during the second half of March, when a large wave of geese and ducks stopped over there. The species diversity was the greatest in the first half of April, when more species arrived from their wintering grounds. We did not find any relationship between water depth and total abundance, but the former did significantly influence the numbers of some ecological feeding groups of waterbirds, i.e. diving omnivores, diving carnivores and surface carnivores. We assume that higher water level in a river valley is favourable to birds during their spring passage, because it creates areas of shallow floodwater that are intensively used by many different aquatic species. Owing to the well-preserved hydrological conditions and its full protection as a national park, the Biebrza valley not only offers optimal conditions for migratory waterbirds, but is also a good research area for studies of the phenomena taking place in avian communities during their migratory staging.


Introduction
Wetlands are important for waterbirds throughout their life cycle (Weller 1999;BirdLife International 2001;Pavón Jordán et al. 2020). They provide suitable breeding habitats and are equally important as roosting and foraging sites during migration and wintering for many species, including highly endangered and declining ones . Artificial and natural water bodies, as well as temporarily flooded river valleys, are of varying significance for wetland birds (Tomiałojć 1993;Faragó & Hangya 2012).
One key factor appears to be the area of a water body and the water level in it; in river valleys, the latter is less stable and often seasonal. Being associated with floods, the floodplain terrace is thus more attractive to waterbirds during overbank flow. Temporary pools provide birds with convenient feeding sites and various potential food sources, such as aquatic vegetation, seeds and invertebrates. They are also safer as roosting areas that are less accessible to people as well as some other predators, e.g. Domestic Dogs (Canis familiaris) and Red Foxes (Vulpes vulpes) (e.g. Jankowiak et al. 2015;Johnston-González & Abril 2019).
Whereas the water level in artificial reservoirs is usually more stable and largely under human management (Zhang et al. 2021), in river valleys it is more variable and depends, among other things, on weather conditions and habitat quality of the valley (Rojas et al. 2011;Faragó & Hangya 2012;Madsen et al. 2014). Water levels are changeable and influenced by such factors as global warming, which may reduce the area flooded because of the smaller amount of ice and snow cover during the preceding winter (Cao et al. 2002;Schneider et al. 2011). Moreover, flood level depend on human interference, as river channel regulation and floodplain drainage, which currently often take place on a large scale, appear to be a key factor in water level decline (Nilsson & Dynesius 1994). In this way, river valleys that used to be important have been destroyed; as a result, there are now fewer refuges, not only for birds, but also for biodiversity in general (Platteeuw et al. 2010;Golawski et al. 2014;Kasprzykowski et al. 2014;Polakowski et al. 2019).
One of the largest contiguous coherent wetland areas and well-preserved river valleys in central Europe is that of the river Biebrza (NE Poland) (Wassen et al. 2006;Grygoruk & Okruszko 2015). The majority of it is protected in the form of the Biebrza National Park, Natura 2000 sites (92/43/ CEE -"Habitats Directive" and "Birds Directive" 2009/147/EC), a Ramsar site and an International Bird Area (BirdLife International 2022). Because of the vast area of wetlands and extensive seasonal flooding in the spring, access to the valley is limited and there is relatively little human disturbance, so wildlife in general, including migratory birds, can benefit from safe conditions (Paquet & Darimont 2010;Jankowiak et al. 2015). The river overflows every year in spring, creating optimal habitats for various waterbirds. It is an important refuge for many breeding and migratory waterbird species, namely Common Crane (Grus grus), Eurasian Wigeon (Mareca penelope), Ruff (Calidris pugnax) and Greater White-fronted Goose (Anser albifrons) (Górski & Nowakowski 1998;Nowakowski 2006;Polakowski et al. 2021;BirdLife International 2022).
In this study we analyzed the influence of water level on bird assemblages during their spring migration in the natural Biebrza valley. We anticipated that the large area of floodwater and their varying depths would be crucial factors for staging waterbirds, as they affect the conditions for foraging and roosting by providing food and a safe haven. Finally, we expected that the abundances of the various ecological (foraging) avian groups would peak at different times of the season, as a potential response to the variable availability of food.

Study area
We carried out the study in the lower part of the Biebrza Basin (north-eastern Poland) (Figure 1). It covered approximately 14 km reach of the river Biebrza, between the villages of Chyliny and Rus, a section that is well preserved at the scale of the whole basin (Grygoruk et al. 2021). The river valley is annually flooded, usually between February and May, though mainly in March and April. This is the spring migration period for waterbirds in this part of Europe. The floodplain can vary in width from 1 to 2 km. In the Lower Biebrza Basin alone, marshes and meadows can become inundated over an average area of 68 km 2 (Chormański 2011), and at the scale of the whole valley, flooded area can vary greatly from one year to another, from 12.5 km 2 to 141.2 km 2 (Ignar et al. 2011). The average depth of the annual flood water at its peak is no more than 0.5 m (Ignar et al. 2011). The research area was divided into 6 sections from 1.4 to 2.3 km in length.

Field study
Fieldwork was performed during 7 consecutive spring seasons in 2015-2021. A total of 69 counts were performed, from 8 to 12 each year, from early February until early May. Wherever possible, all birds were identified to species level. However, because the taxonomic split of bean geese into two species -Tundra Bean Goose (Anser serrirostris) and Taiga Bean Goose (Anser fabalis) (Gill et al. 2022)occurred during the study period, we have adopted the overall grouping Anser fabalis sensu lato to describe the two new species. To avoid differences between observers, the same person (ŁK) always performed the counts. These always took place from the same optimal vantage points at the edge of the valley, that is, the whole area of one section was visible from a single point. Field scopes were used to count birds (Kowa 32x82, Swarovski 25-50 × 65 and 30-70 × 95), and days with good weather were chosen, i.e. no heavy rainfall, poor visibility or very strong winds. On each count day, we also took the river water level reading from the Burzyn water gauge, which is situated in the study area.

Statistical procedures
We modelled the flooding in the surveyed part of the Biebrza valley in HEC-RAS software (Hydrological Engineering Centrer 2021) on the basis of daily flow data from hydrological yearbooks (IMGW 2021). We predicted the extent of flooding in each study year using unsteady flow analysis. We ran the simulation from 1 February to 1 May with the computation intervals set to 2 hours. Then, we created a raster of flooding for each census day at 12:00, thereby obtaining 69 images of flooding. Each pixel was 1 × 1 m in size and had a depth value. This

140
Ł. Krajewski et al. flooding simulation is exemplified by the map for 23 March 2020 (see supplementary materials). Thereafter, we used QGIS software (QGIS.org 2020) to compute the raster statistics. We obtained the depth as the mean of all depth values assigned to each pixel in the six sections. The water surface was shown as a percentage and counted as the number of "water" pixels (simulated water pixels) to the maximum possible pixel number in the given section. The depth variable indicated the average depth assigned to a pixel in a given section during the census period, while the water surface area showed how much of the section was covered by simulated water pixels.

Statistical analysis
When checking for multicollinearity, we found collinearity between day of year and minimum temperature, so we omitted temperature in the subsequent analysis. We found no multicollinearity (VIF < 1.2) for the remaining variables. Before the analysis, the water surface area, depth and temperature were calculated and the variables were centred.
In the species richness analysis, we treated the number of species during the counts as dependent variables. We also modelled the number of species for different ecological (foraging) groups (herbivores, diving omnivores, diving carnivores, shoreline omnivores, surface foraging carnivores). The list of species in each category is given in Table S1 in the supplementary materials; the division into ecological groups is taken from Paszkowski and Tonn (2006). We used mixed models with a negative binomial error distribution and random intercept model structure (random effect structure: year/section; random variation in the intercept among sections within a year and among years). The independent variables were day of year (DOY), depth (mean depth of water) and water surface area (percentage surface of the flooded area). In the second analysis, bird abundance was the dependent variable. All the terms and model structures were the same as above. All analyses were performed in R (R Core Team 2021) statistical software with the MASS package (Venables & Ripley 2002).

Species richness
Day of year was significantly correlated with total species richness (Table I; Figure 2(a)), as well as with species richness in each of the ecological feeding groups (Table I; Figure 2(c, e, g, i, k)). The correlation was always positive, suggesting that species richness increases as the spring progresses.
Water depth was significantly correlated with species richness only in the case of diving omnivores and diving carnivores (Table I; Figure 2(f and h)). It was always positively correlated, i.e. the deeper the water, the greater the richness of species from these two groups. Water surface area was significantly correlated with the total species richness (Table I; Figure 2(b)), as well as with the species richness of herbivores, shoreline omnivores and surface carnivores (Table I; Figure 2(d, j, l)). Again, the correlations were always positive, i.e. the larger the surface area of water, the more species from these three groups were observed.

Abundance of birds
Day of year was significantly correlated with total abundance of birds (Table II; Figure 3(a)), as well as with abundance in each of the ecological feeding groups (Table II; Figure 3(c, e, h, k, m)). In each case the correlation was positive, indicating that

142
Ł. Krajewski et al. total abundance of birds increases with the progress of spring. Water depth was significantly correlated with abundance only in the case of three ecological feeding groups: diving omnivores, diving carnivores and surface carnivores (Table II; Figure 3(g, j and o)). The correlations were positive, i.e. the deeper the water, the more birds from these three groups were observed. Water surface area was significantly correlated with total abun-dance (Table II; Figure 3(b)), as well as with abundance of different ecological feeding groups: herbivores, diving omnivores, shoreline omnivores and surface carnivores (Table II; Figure 3(d, f, l, n)). We found positive correlations for all of them, indicating that the abundance is the larger, the greater the area flooded. In contrast, we obtained a significant negative relationship for abundance of diving carnivores (Figure 3(i)).

Discussion
During the study period, bird migrations in the Biebrza valley were characterized by an increasing number of species and abundance of birds as spring progressed, a generally common pattern in the case of waterbirds (Quan et al. 2002;Baar et al. 2008;Liordos 2011). The species diversity was the greatest in the first half of April, when more species arrived from their wintering grounds (Tomiałojć & Stawarczyk 2003). However, we counted the largest numbers of individual birds in the second half of March, when large numbers of geese and ducks appeared. Feeding opportunities play a crucial part as regards the arrival dates and number of bird species (Shariatinajafabadi et al. 2014;Stafford et al. 2014), and the abundance and availability of food resources are important factors affecting the seasonal dynamics of waterbirds (Ma et al. 2010). We presume that nutrients from the River Biebrza promoted the growth of meadows, so a large population of avian herbivores (especially geese) forage on them (Polakowski & Kasprzykowski 2016). Aquatic invertebrates in the mudflats and shallow water also attracted shorebirds, while areas with deeper water were preferred by diving species, foraging on fish and molluscs. Moreover, the availability of both rich and safe feeding areas during the prenesting period may have general and profound effects on subsequent breeding success and other fitness-related measures, such as survival (Arzel et al. 2006;Polakowski et al. 2019). The extensive Biebrza valley with its diversity of habitats potentially provides such conditions for the entire group of waterbirds owing due to the presence of large wetland complexes containing a variety of different wetland types and a great diversity of food items (2010). Water surface area and depth appear to be key factors affecting waterbird abundance, and many papers indicated the negative impact of higher water levels, as in such conditions some microhabitats, including gravel banks, paved riverbanks,  (Li et al. 2013(Li et al. , 2019Sesser et al. 2016;Aharon-Rotman et al. 2017;Golet et al. 2018;Zhang et al. 2021). The Biebrza valley is a natural floodplain situated in undulating terrain. Even when the water level was at its highest, and a substantial part of the floodplain was covered by shallow water, there were islets of dry ground protruding above the surface. Shallow floodwaters of this kind in the valley are of great importance to waterbird communities, as they provide suitable foraging habitat for many species (Colwell & Taft 2000;Zhang et al. 2021). Variable water levels in a natural river valley offer optimal foraging and roosting conditions for various species. Deeper waters cover only the riverbed and oxbow lakes. Birds can feed on meadows (e.g. Anser sp., Cygnus sp.), shallow flooded areas (e.g. Anas sp., Mareca sp.), as well as on those covered with deeper water (e.g. Goldeneye Bucephala clangula, Cormorant Phalacrocorax carbo). As the spring progresses and floodwaters recede, littoral zones become more exposed, making suitable foraging areas available for waders like Lapwing (Vanellus vanellus) and Ruff. These areas are also used as foraging and roosting sites by Black-headed Gull (Chroicocephalus ridibundus), which formed the core of the bird assemblage during the study period. Furthermore, depending on the foraging group, waterbirds respond differently to water surface area and depth, as the availability of a habitat differs for each type of foraging guild due to its distinctive morphology and techniques that birds use to obtain food (Chatterjee et al. 2020). In line with our expectations, deeper water mediated the availability of areas for diving birds, because they require a minimal water depth in which to forage (Ma et al. 2010). On the other hand, shoreline omnivores depended on shallow water and the edges of meadows with patches of mud, which are essential for their survival (Zhang et al. 2021). The area of such habitats increased along with the decrease in overall water surface area, i.e. the slowly falling water level uncovered habitats for this group of birds. Herbivores, the most numerous group of birds, were likewise very dependent on the surface area of water. Generally, high water levels in the natural valley of the Biebrza were favourable to greater species richness and abundance of birds, because only then did the valley offer them a multitude of feeding habitats. On the other hand, a large area of deeper water could provide only a few species with suitable habitats to feed effectively.
Furthermore, extensive inundated areas are important as feeding grounds, but they also play a crucial role as safe roosts, which may even be decisive for the large assemblages of birds observed here. For example, Greater White-fronted Geese usually stop here for several days during spring passage, when they make use of both the foraging and roosting areas in the valley (Polakowski et al. 2019). Open habitats in the valley are up to 5 km wide in some places. It is likely that these extensive areas of floodwater significantly reduce pressure from predators (both carnivorous mammals and humans), thus providing safe havens for birds (Johnston-González & Abril 2019;van der Kolk et al. 2022).
The results obtained in the natural valley of the river Biebrza are a strong argument in favor of wetland restoration -areas that historically supported a natural ecosystem, but later came to be modified and used for other purposes (Ma et al. 2010). River channels and valleys are being restored in many places worldwide, bringing numerous benefits, including the reduced risk of flooding in developed areas and improved living conditions for animals (e.g. Pedersen et al. 2007;Cui et al. 2009;Jenkins et al. 2010;Hagy et al. 2017). Our results indicate that higher water level in a river valley is favourable for birds during their spring passage. However, we would like to emphasize clearly that in the case of a natural river valley, like that of the Biebrza, even when the water level is high, there are still shallow patches of floodwater, with islets of meadows protruding above the water surface, thus providing many different species with a variety of favourable conditions. This is therefore not the same as the wholesale inundation of an area by deep water, when all bird habitats are lost (Faragó & Hangya 2012;Jankowiak & Ławicki 2014).
To conclude: well-preserved habitats in the Biebrza valley, in conjunction with the natural hydrological regime described in this paper, could be a good example of successful conservation efforts in the management of natural river valleys, which are becoming increasingly rare at the global scale. Such management should be taken into account in projects designed aimed the rehabilitation of river valleys that have been modified and their hydrological regimes distorted. The unique natural values of the Biebrza valley have been preserved, above all, because no drainage work has been undertaken in this area, but also as a result of protection in the national park (Wassen et al. 2006). Therefore, it is a good research area to observe and analyze present-day relationships between hydrological conditions and birds.