Seasonal and Spatial Patterns of Bird Communities in a Highly Disturbed Atlantic Riparian Corridor

Abstract

Land use changes pose major threats to ecosystems, particularly affecting vulnerable habitats, such as riparian forests. These transitional habitats play a crucial role in supporting biodiversity, particularly avian communities. Despite their recognised importance, studies on the land use effects on bird communities in the riparian corridors of southern Europe remain scarce. Here, we aimed to investigate the seasonal variation of the effects of land use on avian communities in an Atlantic riparian area in northern Portugal and whether bird assemblages can be used as bioindicators of riparian ecosystems’ quality. To achieve this, we conducted bird surveys during three periods of the birds’ life cycle: post-nuptial migration, wintering and breeding. Bird species were grouped into assemblages reflecting diet, foraging stratum, phenology and preferred habitat affinities. To analyse the effect of land use, we modelled the abundance of the respective bird assemblages with the land use gradients obtained through principal component analysis. A total of 62 bird species were identified (73% observed during post-breeding migration, 77% in winter and 68% during breeding). Among these, 45 species (73%) were residents, while 17 species (27%) were short- or long-distance migrants. All bird assemblages showed seasonal differences in species richness, with the exception of granivores, forest species, resident, ground- and understorey foragers, and in abundance, with the exception of invertivores, farmland birds and tree foragers. The predicted abundances of farmland birds, ground-feeding birds and granivores often showed positive associations with gradients reflecting anthropogenic land uses (e.g., farmlands and acacia stands) and negative relationships with natural land uses (e.g., deciduous riparian forests, pine and oakwood). Conversely, invertivores’ and tree foragers’ abundances were positively related to natural land uses and negatively related to anthropogenic ones. Furthermore, we highlight the negative effects of exotic tree species on the bird community and the effects caused by adjacent land uses on riparian habitats. Our results are consistent with studies showing that the grouping of birds by functional characteristics can serve as an indicator of disturbance in riparian corridors.

1. Introduction

Anthropogenic pressure is a major driver of ecosystem degradation, affecting its structure and functioning by modifying biological and physicochemical processes [1,2,3]. Highly dependent on natural resources, in response to continued development and population growth, humanity is increasingly putting pressure on the biosphere and exploiting various ecosystem services [4]. For centuries, humans have transformed natural landscapes into various land use activities, ranging from farmland production and forestry to urbanisation, to surpass their essential needs [4]. Such a situation led to the intensification of land use and land cover changes (LULCC), the over-explorative and intensive conversion of one cover type to another and its consequent exploitation by human activities to acquire natural resources and fulfil human needs [5,6,7].

Despite land use activities being essential for humanity, their harmful effects on ecosystems are well known, primarily causing ecosystem degradation and biodiversity decline. Since the 19th century, LULCC have been responsible for approximately 35% of CO₂ emissions into the atmosphere [8], with impacts on climate from local to global scale [4,9,10]. The transformation of natural ecosystems into agriculture or urban areas through deforestation is among the reasons for carbon cycle modification and emissions [5,8]. Another negative effect of LULCC includes, for example, the deterioration of water courses caused by urbanisation or nutrient and agrochemical pollution by agriculture intensification, infectious diseases propagation and soil degradation [4,11].

Riparian zones are transitional areas between the terrestrial and freshwater ecosystems of great ecological importance, balancing temperature and providing organic matter and nutrients for the great diversity they harbour [12,13]. They can act as buffer zones from the effect of habitat fragmentation and mitigate the harmful impacts of exotic woody species from the surroundings [14,15,16]. Similarly, riparian forests are considered highly diverse, dynamic and complex biophysical habitats that function as interfaces between terrestrial and aquatic systems, encompassing sharp environmental gradients, ecological processes and communities [17]. Because of this, they may contribute significantly to landscape connectivity, facilitating dispersal and migration of forest-dependent bird species [18], acting as corridors. The structural complexity of riparian forests, including diverse vegetation layers and proximity to water, enhances habitat quality, making these areas key biodiversity hotspots [19,20]. However, riparian habitats are vulnerable to perturbation and are progressively threatened not only by water pollution but also by land use changes and occupation by invasive alien species [21,22,23].

Recent studies highlight the essential role of riparian corridors in supporting biodiversity and avian communities across the globe, from tropical [24] to temperate ecosystems, such as those found in Europe [25,26]. Riparian corridors provide critical resources, such as nesting sites, food and stopover habitats during migration, supporting diverse bird assemblages [27]. In southern Europe, riparian forest systems act as vital refuges for forest-breeding birds, especially in fragmented agricultural areas, where they sustain higher species richness and abundance compared to surrounding habitats, favouring, for instance, species of conservation concern [28,29,30]. Since birds are usually more responsive to disturbances to their habitats than other cohabitating organisms [31], conservation and restoration of riparian habitats are thus critical strategies for maintaining avian diversity and ensuring the ecological integrity of European landscapes [32,33,34]. Also, their ecology and connection with the vegetation and landscapes are, in general, well known [35,36]; they are often more conspicuous and easier to survey and study than other taxa [37], thus being commonly used as bioindicators of disturbance in riparian corridors.

The objectives of the study are (i) to characterise the bird community in the riparian corridors along two very disturbed rivers in a largely urbanised area of northern Portugal throughout the year (wintering season, breeding season and post-breeding migration period); (ii) to understand the importance of riparian corridors as breeding, wintering and migratory stopover habitats for different bird assemblages; (iii) to determine the effects of land use on the composition and abundance of riparian bird community; (iv) to evaluate the usage of bird assemblages as indicators of riparian ecosystem health. Spatial and temporal variations in bird community were examined, and we also analysed the abundance of specific assemblages reflecting trophic, habitat preference, phenology and foraging substratum affinities to understand how they were affected by different land uses. We hypothesise that riparian corridors have a particular importance for migratory birds, that birds that feed on plant material are related to anthropogenic and perturbed landscapes and that invertivore (insectivores sensu lato, i.e., those that feed on invertebrates) abundance serves as a good ecological state indicator of riparian areas [38,39]; thus, it will likely increase in areas composed mostly of natural habitats (e.g., native deciduous riparian forests), whereas areas dominated predominantly by exotic plant species will negatively impact bird communities.

2. Methodology

2.1. Study Area

We conducted this study in the riparian corridors of two rivers in the municipality of Gondomar, northwestern Portugal (41.1396° N, 8.5291° W, Porto district) (Figure 1): the Sousa River, a tributary of one of Portugal’s largest rivers, the Douro River, with a total extension of 65 km, and the Ferreira River, a tributary of the former with a 43 km extension [40]. In the study area, both rivers are characterised by their sinuous course that lies, for the most part, in low-altitude but pronounced valleys and crosses mostly schist outcrops. The climate is temperate due to Atlantic influence [41], with rainy winters, persistent fogs and dry although not particularly warm summers (between 23 °C and 29 °C) [40].

In the sloping areas, the vegetation cover is dominated by scrubland of heath (Erica spp.), gorse (Ulex spp.), prickled broom (Pterospartum tridentatum), Portuguese broom (Cytisus striatus), and the tree cover is mainly composed of maritime pine (Pinus pinaster) and high densities of the exotic common eucalyptus (Eucalyptus globulus) and invasive acacia trees (Acacia spp.). The riparian forest, on the other hand, although fragmented and taken over by invasive alien species, such as wattle trees (Acacia dealbata and Acacia melanoxylon), or replaced by poplar (Populus spp.) plantations, still contains typical native species in the most undisturbed areas. It consists mainly of common oak (Quercus robur), black alder (Alnus lusitanica), bay laurel (Laurus nobilis), narrow-leaved ash (Fraxinus angustifolia), elder (Sambucus nigra) and willows (Salix atrocinerea and Salix alba) as the most representative woody vegetation, while brambles (Rubus spp.) dominate the shrubby layer.

The deterioration of the riparian ecosystem at the study site is strongly linked to current and former land uses. Agriculture and forestry are the predominant land uses, with corn crops, subsistence agriculture and eucalyptus, pine and poplar planted forests being commonly found along the riverside and surrounding areas [40]. Water quality is poor due to anthropogenic pressure, and in some stretches of both rivers, the artificialisation of the banks is visible, with the presence of schist walls made in the past by the local population in order to prevent flood damage to their farming land, as well as old trails meant to transport materials from mining activities [40]. Moreover, the construction of infrastructure (e.g., roads and urban settlements) excessively close to the riverbanks led to the fragmentation and deterioration of the riparian forest.

2.2. Sampling Design

We selected 50 sampling sites, 29 located on the banks of the Sousa River and 21 in the Ferreira River, along 8.4 km and 6.8 km stretches of each river, respectively (Figure 1). Sampling sites were selected using Google Earth Pro [42] by determining the closest locations to the riverbanks while ensuring a minimum distance of 200 m between them to minimise repetitive encounters [37]. Prior to the surveys, site accessibility was checked on site. Once these requirements were met, all 50 initial points were approved, with only minor rectifications being necessary. The distance between points ranged from 200 to 665 m. After the first bird survey in autumn 2021, two sampling sites on the Ferreira River had to be replaced due to unexpected access restrictions.

2.3. Bird Sampling

In each sampling site, we conducted bird surveys using 50 m radius point counts. Counts were performed within the first four hours after sunrise and under good weather conditions (absence of rain, dense fog or strong winds) to maximise bird contact [37]. During a 10 min period, every individual from all species, visually and audibly identified within the defined radius, was counted [37,43]. Birds were sampled once during the post-breeding migration period (19–23 September 2021), but we conducted two different counts to account for both the wintering season (2–15 December 2021 and 17–24 January 2022) and breeding season (16–21 April 2022 and 17–25 May 2022).

To aid in the interpretation of ecological effects, we categorised bird species’ functional traits according to their diet, phenology, habitat preference and foraging stratum affinities (see Table S2). For the bird assemblages reflecting the diet and foraging stratum, we followed the dataset from Wilman et al. (2014) [44], where we considered the highest percentage of the diet and foraging category used by a bird species. For some categories, namely the foraging stratum, we felt the need to make adaptations, such as merging “midstorey” and “canopy” to form a broader category (“tree foragers”). Habitat preferences were classified as farmland and forest according to the European Bird Census Council (EBCC) species classification [45]. Birds that did not belong to one of those habitats or simply were habitat generalists were classified in the “other” category. Finally, the phenology status was based on Svensson’s established classification for mainland Portugal [46], although the status of some species was adjusted at local level. For the bird checklist, this study followed the taxonomy used by BirdLife International [47]. For the full list of species in Portugal for the breeding, wintering and post-breeding migration periods, we consulted the respective atlas [48,49].

2.4. Habitat Characterisation

Habitat characterisation was conducted from a 100 m buffer around each of the survey points using a land use map (COS 2018) [50], where we mapped all the prominent land covers (see Table S1). We used a buffer larger than the one used for the bird surveys (50 m) to evaluate how the adjacent upland landscape elements influence riparian bird communities. Mapping was refined with information from Google Earth Pro and field work.

We ended up with ten land use categories that were considered for further analysis (Figure 1): urban areas, including both urban settlements and infrastructure (e.g., roads); farmlands, mainly consisting of corn plantations and rotative crops, grasslands and vineyards; shrublands, typically of Ulex spp., Erica spp. and Cytisus striatus associations on the valley slopes; pine forests, with a significant shrubby layer; oak forests (Quercus robur and Quercus pyrenaica), frequently associated with a well-structured and natural understorey; native deciduous riparian forest, majorly represented by Salix atrocinerea, Salix alba, Alnus lusitanica and Fraxinus angustifolia, with a developed understorey layer in undisturbed areas; poplar plantations (hybrids of exotic Populus spp.), typically on riverbanks; eucalyptus forests, either plantations or invasively established; patches of invasive Acacia wattle trees (principally Acacia dealbata and Acacia melanoxylon); abandoned farmlands, former arable and cultivated areas now occupied essentially by shrubs (particularly Rubus spp.), herbaceous species and dispersed younger stages of the most abundant tree species. We excluded the cover by “water bodies” because it was nearly the same in all the survey points.

We calculated the proportion of each land use cover per sampling point on QGIS (v3.22.7) [51], to use as variables in the statistical analysis.

2.5. Data Analysis

For each season, we used the maximum number of each species observed in each sampling site during the two sampling periods. With this approach, we accounted for both residents that may be more active during early and mid-spring, as well as the late arrival of long-distance migrants, usually from mid-March onwards, which results in the potential for breeding activity to be extended until late spring. Then, we described the overall patterns of bird assemblages of the Ferreira and Sousa Rivers during each season (breeding, post-breeding migration and wintering) using species richness, occurrence and mean abundance (Table S3).

We used one-way ANOVA to examine the seasonal variations in the abundance and species richness of bird functional groups. We tested for differences in mean cover of the main land uses between the Sousa and Ferreira Rivers using a two-sample t-test.

To reduce dimensionality and simplify the determination of the dominant gradients in the landscape [52], we performed a principal component analysis (PCA) on the 10 land use categories in all sampling sites using the “prcomp” function. A varimax rotation on components with eigenvalues > 1 was applied. The resulting rotated axes were further considered as variables in the modelling process.

We used generalized linear models (GLMs) with a Poisson family and log-link transformation [53] to analyse the land use effects within the riparian corridor on the abundance of each bird assemblage and each season. The model terms were based on the previously obtained PCA rotated axes and the river variable (Sousa or Ferreira). We used a multi-model inference (MI) approach to assess individual variable importance, which is a method based on an estimated weighted average across all subset models containing a specific variable that takes observed model weights into account [54]. We further obtained an average model, based on all model combinations generated, which contained the variables with an average estimate of all coefficients.

All statistical analyses were performed using R 4.2.1 software [55]: “dplyr” package [56] for the ANOVA, “psych” package [57] for the PCA and “MuMIn” package (Bartoń, 2016) [58] for the modelling process.

3. Results

3.1. Overall Bird Patterns

Altogether, 62 bird species were identified, of which 73% occurred in the post-breeding migration period, 77% in winter and 68% in the breeding season. Of the 62 observed species, 45 (73%) were resident, and 17 (27%) were either short- or long-distance migrant species (see Table S2). During the spring sampling period, this study identified 42 breeding bird species, corresponding to ca. 18.5% of the number of breeding species for mainland Portugal, while for the wintering and post-breeding migration communities, we identified 61 species, which roughly corresponded to 15% of the total birds recorded in Portugal in the same period. Regarding their habitat preferences, habitat generalists represented 45% of total birds, while forest birds represented 35% and farmland birds 20% of the bird community. Invertivores (45% overall) strongly predominated, being the most represented diet group in all seasons (34% in winter, 42% and 44% in the breeding and post-breeding migration seasons, respectively), while ground foragers (40%) also predominated all year round (migration season: 38%; wintering season: 42%; breeding season: 48%) relative to the foraging stratum classes.

In the post-breeding migration season, the most frequent and abundant species in the study area were the European robin (Erithacus rubecula) (72%; 1.5 ± 1.4), European pied flycatcher (Ficedula hypoleuca) (54%; 0.8 ± 1.0) and Cettia’s warbler (Cettia cetti) (54%; 0.7 ± 0.7).

The European robin was the most frequent species on the rivers’ riparian corridors during winter (96%) and the post-breeding migration season (72%). In the winter, apart from the European robin, which occurred in 96% of the sampling sites, the most frequent species were the common chaffinch (Fringilla coelebs) (84%), the blackbird (Turdus merula) (84%), the northern wren (Troglodytes troglodytes) (84%), the common chiffchaff (Phylloscopus collybita) (68%), the great cormorant (Phalacrocorax carbo) (66%), Cetti’s warbler (Cettia cetti) (66%) and the Eurasian blackcap (Sylvia atricapilla) (64%). The most abundant species during the winter were the common chaffinch (4.8 ± 9.4), the Eurasian siskin (Spinus spinus) (4.7 ± 17.7), the European robin (2.5 ± 1.3) and the European serin (Serinus serinus) (2.2 ± 8.9) (see Table S3).

During the breeding season, the most frequent species on the rivers’ riparian corridors was the northern wren (98%), the Eurasian blackcap (90%), the common firecrest (Regulus ignicapilla) (74%), the blackbird (74%), the great tit (Parus major) (66%) and the European robin (64%). The most abundant species in the breeding season were the European robin (1.0 ± 0.9), the common firecrest (1.0 ± 0.9), the European serin (1.2 ± 1.6), the blackbird (1.3 ± 1.1), the northern wren (1.9 ± 0.9) and the Eurasian blackcap (1.9 ± 1.0) (see Table S3).

There were significant differences in the overall bird community in the Sousa and Ferreira Rivers between the seasons for all species, invertivores, omnivores, carnivores, farmland species, other habitat species, resident species, as well as for species that feed on trees, on water and unspecialised species (

Table 1. One-way ANOVA results for each bird assemblage abundance and species richness during the three different seasons. Significant relations (p < 0.05) are highlighted in bold.

Bird Assemblage	Abundance		Species Richness
	F	p-Value	F	p-Value
All species	18.75	<0.001	3.42	0.04
Diet
Invertivore	2.57	0.08	3.36	0.04
Omnivore	22.21	<0.001	3.31	0.04
Granivore	4.96	0.01	1.93	0.15
Carnivore	21.04	<0.001	17.52	<0.001
Habitat
Forest	15.01	<0.001	1.69	0.19
Farmland	1.60	0.21	4.35	0.01
Other	6.74	<0.001	3.37	0.04
Resident	15.23	<0.001	0.12	0.88
Foraging Stratum
Tree	2.47	0.09	5.99	<0.001
Ground	9.63	<0.001	0.28	0.76
Water	5.50	<0.001	14.98	<0.001
Understorey	20.43	<0.001	1.60	0.21
Unspecialised	9.32	<0.001	9.46	<0.001

).

In terms of abundance, the all-species bird assemblage showed significant differences in the number of individuals, as well as omnivores, granivores, carnivores, forest species, other habitat species, resident species, as well as species that feed on the ground, on water, on the understorey layer and unspecialised foragers (Table 1).

Differences in the bird community between the rivers were only significant during the wintering season (see Tables S4–S6). Specifically, the Ferreira River had higher abundances of farmland birds (Ferreira: 8.61; Sousa: 0.90; t-test = 2.10; p-value = 0.04), ground-foraging birds (Ferreira: 17.91; Sousa: 7.00; t-test = 2.67; p-value = 0.01) and resident species (Ferreira: 28.3; Sousa: 17.1; t-test = 2.21; p-value = 0.04).

3.2. Habitat Patterns

The study comprised 10 main land use types (Figure 1,

Table 2. Two-sample t-test for mean proportions of cover by land uses in the Ferreira and Sousa Rivers.

Land Use	Mean Ferreira	Mean Sousa	t-Statistics	df	p-Value	95% Conf. Int.
Abandoned farmland	0.06	0.08	−0.50	50	0.62	−0.08	0.04
Acacia	0.00	0.03	−2.08	50	0.04	−0.04	0.00
Deciduous forest	0.17	0.26	−2.14	50	0.04	−0.17	−0.01
Eucalyptus	0.12	0.16	−1.33	50	0.19	−0.11	0.02
Farmland	0.22	0.14	1.60	50	0.12	−0.02	0.17
Oak forest	0.04	0.03	0.36	50	0.72	−0.03	0.04
Pine forest	0.06	0.05	0.63	50	0.53	−0.04	0.07
Poplar plantation	0.04	0.02	1.01	50	0.32	−0.01	0.04
Shrubland	0.07	0.05	0.89	50	0.38	−0.03	0.07
Urban	0.14	0.08	1.79	50	0.08	−0.01	0.14

), showing a heterogenous riparian corridor and surrounding landscape. Urban areas were the most prevalent land use category in the Ferreira River (91%), while farmland had the highest mean proportions (mean = 0.22). In the Sousa River, on the other hand, deciduous riparian forest was the most recurrent type of land use (96.6%) and had the highest proportion (mean = 0.26). There were no major differences between the two sampled rivers, although the cover by Acacia and deciduous forests was slightly higher in the Sousa than in the Ferreira River (p-value = 0.04) (Table 2).

The PCA on the ten variables describing land use extracted five rotated axes (

Table 3. Loadings of the land use, as variables, on varimax rotated axes (RPC#) extracted from a principal component analysis (PCA).

Variables	RPC1 1	RPC2 2	RPC3 3	RPC4 4	RPC5 5
Abandoned farmland	0.01	−0.08	−0.05	0.95	0.05
Acacia	0.04	−0.02	0.00	0.04	0.92
Deciduous forest	0.80	−0.05	−0.13	−0.38	0.18
Eucalyptus	0.02	−0.11	0.95	0.01	−0.12
Farmland	−0.75	−0.33	−0.09	−0.41	−0.09
Oak forest	0.22	0.76	0.08	0.04	−0.01
Pine forest	−0.12	0.85	−0.17	−0.03	0.21
Poplar plantation	−0.03	0.68	0.10	−0.10	−0.37
Shrubland	0.75	−0.05	0.24	0.08	−0.11
Urban	−0.39	−0.33	−0.65	0.10	−0.29
Proportion variance	0.20	0.20	0.15	0.13	0.12

Legend: ¹ Deciduous forests and shrublands to farmland; ² Oak, pine and poplar forests; ³ Eucalyptus to urban area; ⁴ Abandoned farmlands; ⁵ Acacia stands. Values in bold indicate |factor loadings| > 0.6.

). The main gradients contrasted sites with dominant cover by deciduous riparian forests connected by upland shrublands over farmland (RPC1; 20%) and increasing cover by oakwoods, pinewoods and poplar stands (RPC2, 20%). Other PCs included RPC3 (15%), representing contrasting increases in cover by Eucalyptus versus urban area; RPC4 (13%), with increasing cover by abandoned farmland; and RPC5 (12%), representing increasing cover by Acacia spp. stands.

3.3. Bird Assemblage in Relation to Habitat Heterogeneity Throughout Different Seasons

3.3.1. Post-Breeding Migration Season

The GLM results showed that the gradient of deciduous forests connected to shrublands over farmland areas (RPC1) had a negative effect on the abundance of farmland species (0.75; p = 0.039) (

Table 4. Generalized linear models (GLMs) relating habitat preference bird assemblage abundance and the rotated axes from PCA analysis and the river variable for each season. The estimated coefficients (Coef), the sum of weights (SoW) of each model and the respective p-value (p) are presented.

Bird Assemblage	Variable	Wintering			Breeding			Migration
		SoW	Coef	p	SoW	Coef	p	SoW	Coef	p
Farmland	RPC1	0.99	−0.99	<0.001	1.00	−0.72	<0.001	0.75	−1.04	0.039
	RPC2	0.32	−0.31	0.138	0.44	−0.24	0.139	0.70	−0.98	0.049
	RPC3	0.32	−0.27	0.245	0.93	−0.42	0.006	0.51	−0.77	0.123
	RPC4	0.23	−0.07	0.733	0.73	−0.34	0.032	0.26	−0.31	0.528
	RPC5	0.29	0.21	0.261	0.35	−0.11	0.570	0.81	1.13	0.033
	River (Sousa)	0.97	−1.70	<0.001	0.22	−0.07	0.886	0.23	0.02	0.975
Forest	RPC1	0.31	−0.10	0.407	0.31	0.05	0.367	0.32	0.06	0.393
	RPC2	0.25	−0.06	0.602	0.59	0.09	0.063	0.98	0.17	<0.001
	RPC3	0.77	−0.26	0.020	0.75	0.11	0.035	0.26	0.03	0.706
	RPC4	0.24	0.04	0.677	0.43	0.07	0.161	0.34	0.06	0.295
	RPC5	0.23	−0.02	0.877	0.68	0.10	0.063	0.69	−0.13	0.046
	River (Sousa)	0.25	−0.11	0.653	0.39	0.13	0.432	0.31	0.11	0.38
Other	RPC1	0.23	<0.001	0.951	0.29	0.04	0.453	0.33	−0.07	0.395
	RPC2	0.24	−0.03	0.737	0.43	−0.08	0.203	0.24	−0.02	0.756
	RPC3	0.24	0.02	0.929	0.26	−0.03	0.612	0.59	−0.12	0.091
	RPC4	0.55	0.13	0.089	0.29	0.04	0.449	0.38	−0.08	0.252
	RPC5	0.24	−0.03	0.650	0.77	−0.14	0.042	0.37	0.08	0.216
	River (Sousa)	0.24	0.05	0.721	0.25	0.02	0.807	0.84	−0.33	0.019

Legend: RPC1—Deciduous riparian forests and shrublands to farmland; RPC2—Oak, pine and poplar forests; RPC3—Eucalyptus to urban area; RPC4—Abandoned farmlands; RPC5—Acacia stands. Both the most suitable models and significant variables (<0.05) are represented in bold. Green = significantly positive; Yellow = significantly negative.

), granivores (0.73; p = 0.023) (

Table 5. Generalized linear models (GLMs) relating diet bird assemblage abundance and the rotated axes from PCA analysis and the river variable for each season. The estimated coefficients (Coef), the sum of weights (SoW) of each model and the respective p-value (p) are presented.

Bird Assemblage	Variable	Wintering			Breeding			Migration
		SoW	Coef	p	SoW	Coef	p	SoW	Coef	p
Invertivore	RPC1	0.32	−0.06	0.382	0.46	0.09	0.164	0.33	0.06	0.390
	RPC2	0.89	0.14	0.009	0.77	0.13	0.018	0.44	0.09	0.143
	RPC3	0.25	−0.03	0.718	0.33	0.06	0.382	0.39	−0.08	0.187
	RPC4	0.95	0.16	0.003	0.60	0.11	0.065	0.23	7.0e−4	0.991
	RPC5	0.24	−0.02	0.571	0.24	0.02	0.876	0.67	−0.15	0.054
	River (Sousa)	0.31	−0.11	0.399	0.46	0.18	0.253	0.26	0.06	0.450
Omnivore	RPC1	0.24	−0.03	0.891	0.33	0.05	0.367	0.24	−0.02	0.813
	RPC2	0.30	−0.08	0.345	0.30	−0.04	0.464	0.70	0.13	0.037
	RPC3	0.43	0.15	0.079	0.74	0.11	0.045	0.92	0.18	0.007
	RPC4	0.37	0.11	0.192	0.44	0.7	0.162	0.58	0.11	0.069
	RPC5	0.24	−0.03	0.891	0.29	0.04	0.551	0.24	0.02	0.630
	River (Sousa)	0.62	−0.34	0.043	0.44	0.15	0.348	0.24	−0.03	0.747
Granivore	RPC1	0.92	−0.59	0.009	0.98	−0.54	<0.001	0.73	−0.59	0.023
	RPC2	0.22	−0.01	0.856	0.36	−0.22	0.161	0.85	−0.84	0.006
	RPC3	0.98	−0.73	<0.001	0.91	−0.45	<0.001	0.98	−1.03	<0.001
	RPC4	0.23	−0.10	0.553	0.78	−0.40	0.014	0.67	−0.55	0.033
	RPC5	0.24	−0.10	0.910	0.28	−0.15	0.606	0.70	0.45	0.034
	River (Sousa)	0.48	−0.67	0.146	0.39	−0.43	0.239	0.25	−0.20	0.622
Carnivore	RPC1	0.24	−0.01	0.972	0.27	0.16	0.585	0.38	−0.21	0.293
	RPC2	0.80	−0.30	0.038	0.36	−0.34	0.372	0.27	0.09	0.592
	RPC3	0.24	0.00	0.999	0.55	−0.43	0.101	0.53	−0.28	0.134
	RPC4	0.26	0.05	0.607	0.37	−0.33	0.404	0.25	−0.07	0.693
	RPC5	0.30	0.08	0.403	0.24	−0.01	0.925	0.25	0.05	0.625
	River (Sousa)	0.24	0.00	0.826	0.38	0.6	0.314	0.49	−0.52	0.219

Legend: RPC1—Deciduous riparian forests and shrublands to farmland; RPC2—Oak, pine and poplar forests; RPC3—Eucalyptus to urban area; RPC4—Abandoned farmlands; RPC5—Acacia stands. Both the most suitable models and significant variables (<0.05) are represented in bold. Green = significantly positive; Yellow = significantly negative.

) and ground-foraging birds (0.98; p < 0.001) (

Table 6. Generalized linear models (GLMs) relating foraging stratum bird assemblage abundance and the rotated axes from PCA analysis and the river variable for each season. The estimated coefficients (Coef), the sum of weights (SoW) of each model and the respective p-value (p) are presented.

Bird Assemblage	Variable	Wintering			Breeding			Migration
		SoW	Coef	p	SoW	Coef	p	SoW	Coef	p
Tree	RPC1	0.62	0.19	0.059	0.45	0.12	0.145	0.37	0.16	0.277
	RPC2	0.70	0.19	0.029	0.33	0.08	0.281	0.42	0.17	0.168
	RPC3	0.67	0.20	0.059	0.86	0.22	0.013	0.26	−0.08	0.568
	RPC4	0.30	0.09	0.371	0.87	0.20	0.008	0.40	0.15	0.210
	RPC5	0.23	0.00	0.886	0.27	0.06	0.463	0.29	−0.12	0.291
	River (Sousa)	0.24	0.01	0.888	0.24	0.05	0.973	0.30	0.23	0.333
Ground	RPC1	0.45	−0.17	0.160	1.00	−0.24	<0.001	0.98	−0.33	<0.001
	RPC2	0.23	−0.01	0.986	0.40	−0.08	0.216	0.95	−0.34	<0.001
	RPC3	0.23	−0.01	0.870	0.99	−0.19	<0.001	0.60	−0.18	0.047
	RPC4	0.24	0.05	0.599	0.79	−0.13	0.027	0.22	−0.02	0.822
	RPC5	0.23	0.02	0.968	0.74	−0.13	0.041	0.75	0.20	0.045
	River (Sousa)	0.99	−0.89	<0.001	0.24	0.04	0.679	0.33	0.23	0.337
Water	RPC1	0.25	0.05	0.724	0.34	0.42	0.303	0.26	−0.11	0.628
	RPC2	0.96	−0.39	0.009	0.39	−0.46	0.344	0.43	0.22	0.198
	RPC3	0.33	0.10	0.391	0.30	0.29	0.679	0.78	−0.45	0.041
	RPC4	0.99	0.28	<0.001	0.24	−0.12	0.762	0.23	−0.01	0.943
	RPC5	0.25	−0.05	0.552	0.31	−0.47	0.240	0.24	0.02	0.763
	River (Sousa)	0.31	0.20	0.391	0.32	0.74	0.476	0.55	−0.65	0.943
Understorey	RPC1	0.24	−0.03	0.872	0.57	0.12	0.087	0.30	0.09	0.327
	RPC2	0.24	−0.01	0.951	0.83	0.15	0.012	0.43	0.13	0.202
	RPC3	0.49	−3.2e−3	0.143	0.26	0.04	0.562	0.25	0.03	0.584
	RPC4	0.47	0.02	0.133	0.55	0.11	0.087	0.24	0.01	0.906
	RPC5	0.53	−0.01	0.125	0.43	0.09	0.219	0.33	0.10	0.215
	River (Sousa)	0.33	0.05	0.639	0.36	0.16	0.501	0.58	−0.37	0.050
Unspecialised	RPC1	0.48	−0.19	0.134	0.26	−0.06	0.615	0.23	−0.06	0.891
	RPC2	0.23	5.0e−4	0.932	0.24	−0.04	0.722	0.44	−0.04	0.140
	RPC3	1.00	−0.49	<0.001	0.30	0.09	0.376	0.27	0.09	0.366
	RPC4	0.23	−0.02	0.873	0.25	0.05	0.645	0.80	−0.33	0.002
	RPC5	0.23	0.01	0.896	0.30	0.09	0.360	0.30	−0.13	0.521
	River (Sousa)	0.23	0.07	0.756	0.24	0.01	0.817	0.48	−0.39	0.169

Legend: RPC1—Deciduous riparian forests and shrublands to farmland; RPC2—Oak, pine and poplar forests; RPC3—Eucalyptus to urban area; RPC4—Abandoned farmlands; RPC5—Acacia stands. Both the most suitable models and significant variables (<0.05) are represented in bold. Green = significantly positive; Yellow = significantly negative.

). This gradient was the most representative for ground foragers.

The pinewood, oakwood and poplar stands gradient (RPC2) reflected a broader cast of associated bird assemblages. As expected, the abundance of forest birds was highly and positively influenced by these land use types (0.98; p < 0.001) (Table 4), being the most relevant model for this bird assemblage. For omnivores (0.70; p = 0.037) (Table 5), the variable showed positive effects on abundance but was not the most important variable in representing the relation between land use and bird assemblage abundance. An opposite effect could be observed for farmland birds (0.70; p = 0.049) (Table 4), granivores (0.85; p = 0.006) (Table 5) and ground-foraging species (0.95; p < 0.001) (Table 6).

Increasing cover by eucalyptus (to the detriment of urban cover) (RPC3) had a positive relation with omnivore abundance (0.92; p = 0.007) (Table 5) and a negative impact on the abundance of granivores (0.98; p < 0.001) (Table 5), ground-feeding birds (0.60; p = 0.047) and water-foraging birds (0.78; p = 0.041) (Table 6), showing an overall pattern of negative effects of eucalyptus stands on the breeding community. This gradient was the most relevant variable that reflected the relation between bird assemblage abundance and the type of land use for all the above-described bird assemblages.

Abandoned farmlands (RPC4) demonstrated significant and negative relations with unspecialised species (0.80; p = 0.002) (Table 6). Acacia tree stands (RPC5) also showed a significant and positive association with the abundance of farmland birds (0.081; p = 0.033) (Table 4) and ground foragers (0.75; p = 0.045) (Table 4) but significantly negative effects on forest species abundance (0.69; p = 0.046) (Table 4). For farmland birds (0.81; p = 0.033) (Table 4), the gradient that represents the presence of acacia trees (RPC5) was the most important variable, whereas for forest species (0.69; p = 0.046) (Table 4) and ground foragers (0.75; p = 0.045) (Table 6), it was also relevant.

Finally, the Sousa and Ferreira Rivers variable was the most important in describing the negative association with other species (0.84; p = 0.019) (Table 4).

3.3.2. Wintering Season

Contrary to the previous season, the results showed an absence of relation between the deciduous forest and shrubland over farmland gradient (RPC1) and the abundance of ground foragers, which were replaced by the all species (0.67; p = 0.070) (

Table 7. Generalized linear models (GLMs) relating phenological bird assemblage abundance and the rotated axes from PCA analysis and the river variable for each season. The estimated coefficients (Coef), the sum of weights (SoW) of each model and the respective p-value (p) are presented. Both the most suitable models and significant variables (<0.05) are represented in bold. Green = significantly positive; Yellow = significantly negative.

Bird Assemblage	Variable	Wintering			Breeding			Migration
		SoW	Coef	p	SoW	Coef	P	SoW	Coef	p
Migrant	RPC1	-	-	-	-	-	-	0.24	0.03	0.925
	RPC2	-	-	-	-	-	-	0.23	−0.01	0.996
	RPC3	-	-	-	-	-	-	0.25	−0.08	0.651
	RPC4	-	-	-	-	-	-	0.34	0.13	0.380
	RPC5	-	-	-	-	-	-	0.26	−0.08	0.547
	River (Sousa)	-	-	-	-	-	-	0.24	0.08	0.729
Non-Breeding	RPC1	0.31	−0.15	0.340	-	-	-	0.32	−0.27	0.380
	RPC2	0.40	−0.22	0.145	-	-	-	0.24	−0.06	0.847
	RPC3	0.99	−0.57	<0.001	-	-	-	0.40	−0.36	0.300
	RPC4	0.23	−0.02	0.974	-	-	-	0.29	0.20	0.405
	RPC5	0.23	0.05	0.756	-	-	-	0.46	0.39	0.076
	River (Sousa)	0.23	0.01	0.674	-	-	-	0.66	−1.22	0.065
Breeding	RPC1	-	-	-	0.24	0.08	0.473	-	-	-
	RPC2	-	-	-	0.31	0.26	0.675	-	-	-
	RPC3	-	-	-	0.33	−0.54	0.321	-	-	-
	RPC4	-	-	-	0.57	0.68	0.097	-	-	-
	RPC5	-	-	-	0.33	−0.99	0.395	-	-	-
	River (Sousa)	-	-	-	0.26	0.54	0.482	-	-	-
All Species	RPC1	0.67	−0.18	0.070	0.27	−0.03	0.495	0.41	−0.09	0.152
	RPC2	0.24	−0.04	0.591	0.24	−0.01	0.884	0.23	0.01	0.840
	RPC3	0.55	−0.17	0.069	0.23	0.00	0.882	0.52	−0.11	0.096
	RPC4	0.27	0.07	0.366	0.24	0.02	0.765	0.24	−0.02	0.803
	RPC5	0.23	−0.01	0.926	0.24	−0.01	0.750	0.28	0.05	0.434
	River (Sousa)	0.45	−0.28	0.171	0.25	0.05	0.573	0.26	−0.07	0.668
Resident	RPC1	0.51	0.14	0.128	0.27	−0.03	0.481	0.51	−0.11	0.115
	RPC2	0.23	0.03	0.727	0.24	−0.01	0.853	0.23	0.01	0.769
	RPC3	0.23	0.02	0.836	0.23	6.1e−4	0.891	0.46	−0.11	0.146
	RPC4	0.37	0.11	0.190	0.24	0.01	0.816	0.29	−0.06	0.443
	RPC5	0.23	2.7e−3	0.961	0.24	−0.01	0.794	0.31	0.06	0.353
	River (Sousa)	0.88	−0.48	0.016	0.25	0.05	0.577	0.26	−0.07	0.694

Legend: RPC1—Deciduous riparian forests and shrublands to farmland; RPC2—Oak, pine and poplar forests; RPC3—Eucalyptus to urban area; RPC4—Abandoned farmlands; RPC5—Acacia stands.

) and species foraging in trees bird assemblages (0.62; p = 0.059) (Table 6). The all-species bird assemblage had a negative relation with riparian deciduous forests and shrublands but a positive relation with farmland (Table 4), whereas tree-foraging species had a positive relation with deciduous forests and shrublands and a negative relation with farmlands. This variable continued to be the most statistically significant and relevant in representing the negative relation of this habitat gradient with the abundance of farmland (0.99; p < 0.001) (Table 4) and granivore species (0.92; p = 0.009) (Table 5) during the winter.

Also, the oakwoods, pinewoods and poplar stands gradient (RPC2) was important in explaining the relation with the abundance of bird assemblages that were not represented in the previous season. It was important to explain the negative relation with the abundance of carnivores (0.80; p = 0.038) (Table 5), being statistically significant and positively related to invertivores (0.89; p = 0.009) (Table 5) and statistically significant and negatively related to species that forage on water (0.96; p = 0.009) (Table 6). This gradient was also the most representative variable (0.70; p = 0.029) and was positively associated with tree species abundance (Table 6).

The eucalyptus to the detriment of urban area gradient (RPC3) continued to be reflected as the most important variable for granivores (0.98; p < 0.001) (Table 5) and unspecialised foraging species (1.00; p < 0.001) (Table 6), with both groups negatively relating with the gradient. Forest (0.77; p = 0.020) (Table 4) and non-breeding birds’ abundance (0.99; p < 0.001) were also negatively related to and mainly affected by this gradient (Table 7).

The presence of abandoned farmlands gradient (RPC4) was shown to be the most relevant variable in representing the relation between the type of land use and the abundance of invertivores (0.95; p = 0.003) (Table 5) and species that forage on water (0.99; p < 0.001) (Table 6). The GLM model revealed positive relations between abandoned farmlands and invertivores and water-feeding species.

The river variable was associated with distinct bird assemblages compared to the post- breeding migration season. It showed significant effects and major importance for residents (0.88; p = 0.016) (Table 7) and ground-foraging species (0.99; p < 0.001) (Table 6). It was also the most important variable for omnivores (0.62; p = 0.043) (Table 5) and showed high statistical significance, but not as the most representative variable, for farmland species (0.97; p < 0.001) (Table 4). The abundance of all these bird assemblages had a negative relation with the river variable.

3.3.3. Breeding Season

The gradient associated with deciduous forest and shrubland over farmland (RPC1) remained important for farmland species (1.00; p < 0.001) (Table 4) and granivore abundance (0.98; p < 0.001) (Table 5). This gradient was the most important variable for ground-foraging species (1.00; p < 0.001) (Table 6), as it related to winter and the post-breeding migration period. For this season, the abundance of all bird assemblages maintained the same relationship as in the previous seasons, i.e., it was negatively related to deciduous forests and shrublands but positively related to farmlands.

For the oak, pine and poplar forests variable (RPC2), the results demonstrated statistical significance for the abundance of invertivores (0.77; p = 0.018) (Table 5) and understorey-foraging species (0.83; p = 0.012) (Table 6), as well as a positive relation with the gradient.

The model demonstrated that the eucalyptus versus urban area variable (RPC3) was significant and had positive effects on the abundance of forest species (0.75; p = 0.035) (Table 4), omnivores (0.74, p = 0.045) (Table 5) and tree-foraging species (0.86; p = 0.013) (Table 6). Additionally, it was the major explanatory variable for these bird assemblages, with the exception of tree-foraging species. The results showed statistical significance and a negative impact for farmland species (0.93; p = 0.006) (Table 4), granivores (0.91; p < 0.001) (Table 5) and ground-foraging species (0.99; p < 0.001) (Table 6). The abundance of granivores, farmland birds and ground-foraging species demonstrated a negative relation with eucalyptus forests and a positive relation with urban areas. Omnivores, forest birds and tree-foraging species showed an opposite relation compared with the previous bird assemblages, where the gradient positively affected the abundance of these bird assemblages.

Throughout all seasons, the abandoned farmland variable (RPC4) had various numbers of bird assemblages associated with it. In the breeding season, it was the most important variable for the abundance of tree-foraging birds (0.87; p = 0.008) (Table 6). For farmland birds (0.73; p = 0.032) (Table 4), granivores (0.78; p = 0.014) (Table 5) and ground-foraging species (0.79; p = 0.027) (Table 6), abandoned farmlands were relevant, but they were not statistically significant for invertivores (0.60; p = 0.065) (Table 5) during the breeding season. Invertivores, breeding birds and tree-foraging species showed a positive relation with this variable, while granivores, farmland birds and ground-foraging species presented a negative relation with it.

The presence of acacia woods (RPC5) was shown to be the most important variable for other species (0.77; p = 0.042) (Table 4) and ground-foraging birds (0.74; p = 0.041) (Table 6). This variable was also relevant for the abundance of forest birds, although it was not statistically significant (0.68; p = 0.63) (Table 4). The acacia gradient had a negative relation with the other species bird assemblage and ground-foraging birds, and it positively affected the abundance of forest birds.

4. Discussion

This study underpins the seasonal importance of riparian forest corridors for avian communities. Results suggest that Iberian riverine habitats may play an important role as stopover sites for the autumnal Afro-Western Palearctic migrant flyway, with several long-distance migrants using this region’s riparian forest ecosystems. They are also considerably important for short-distance migrant species, which use the area as a wintering ground. Additionally, this study highlights the importance of different land uses for avian community composition, as the bird assemblages’ abundance varied greatly across the riparian landscape, suggesting an overall detrimental impact of invasive alien tree species on the avian community. Pine, oak and poplar tree planted forests had several positive associations with a range of bird assemblage abundances in different seasons, namely forest birds, omnivores, tree and understorey foragers and invertivores. Contrary to our expectations, we noticed that the deciduous riparian forest when associated with shrubland had no significantly positive relation with any bird assemblage. This study also consolidates the view that certain avian assemblage groups such as invertivores, granivores, ground foragers and both forest and farmland birds, can be used as bioindicators of deterioration in riparian landscapes. This was evidenced in our study by the negative associations of these bird assemblages with anthropogenic land uses and positive associations with natural land covers.

The bird community in our study area was typical of riverine habitats in the Atlantic region [59] such as the common kingfisher (Alcedo atthis) or the grey wagtail (Motacilla cinerea), with the presence of species associated with other habitats, such as shrublands (e.g., the Dartford warbler (Curruca undata) and the rock bunting (Emberiza cia)) due to natural or human-induced habitat heterogeneity of land uses adjacent to the water course. However, species typical of this geographic region but scarcer and with stricter habitat requirements, such as the white-throated dipper (Cinclus cinclus) [59], were absent. The presence/absence of such species can serve as a bioindicator of water and habitat quality, since they occupy high trophic levels on aquatic food webs and are often highly sensitive to environmental changes [60,61,62].

The bird community of the Ferreira and Sousa Rivers is predominantly dominated by resident species, while migrants constitute 27% of the overall community. Species richness was higher in the wintering and post-breeding migration seasons than in the breeding season, possibly suggesting a relatively impoverished breeding community, which may be linked with habitat disturbance. In fact, there are some studies associating the decrease in the number of species in the riparian corridors with the deterioration (or even absence) of riparian vegetation across the water lines [63], particularly in breeding passerine populations [64]. We found significant seasonal differences in the number of species for all bird assemblages, with the exception of granivores, forest birds, residents, ground foragers and understorey foragers, and in abundance, with the exception of invertivores, tree foragers and farmland birds. As nesting sites, both rivers host lower numbers of species compared to the other two seasons. The breeding community mostly comprised residents and common species, with a scarcity or even absence of breeding migrants, such as the common cuckoo (Cuculus canorus) or the Iberian chiffchaff (Phylloscopus ibericus). This is in line with other studies in other European riparian galleries [25,38,65]. Forest birds (e.g., the firecrest, the great tit, the long-tailed tit and the song thrush) and understorey specialist bird species (e.g., the blackcap and the wren) dominated the breeding community. The verified seasonal pattern of the community reflects the dynamics of different bird assemblages and how they respond to the particularities of each season. Some of the species included as residents actually have short- and long-distance migratory individuals in their wintering and post-breeding migratory populations. In the winter, the European robin, the common chaffinch and the song thrush reached their peak abundance as a result of the migration of individuals from higher latitudes [66], while exclusively wintering species, such as the chiffchaff and the siskin, were also frequent and abundant. This high influx of wintering birds occurs because such species find mild weather conditions and greater availability of resources to survive the winter in the Iberian Peninsula [67,68]. The humid and milder conditions of winter in this region under the influence of the Atlantic climate boost ecosystem productivity and as a result increase the availability of invertivores and seeds, which are crucial food resources for invertivore, granivore and omnivore species [67,68,69]. Also, during the migratory passage, both species from northern Europe and species returning to their African wintering grounds cohabit the riparian corridors of both rivers. Moreover, the occurrence of more specific habitat species, such as the purple heron (Ardea purpurea) or the common reed warbler (Acrocephalus scirpaceus), typically associated with wetlands and dense reedbeds, respectively [70], exemplifies the importance of riparian corridors as migratory stopover sites for birds that must fulfil the energy demands of migration [71].

Our results are in line with other studies suggesting that invertivores and tree foragers are associated with undisturbed riparian areas, and birds feeding on plant material point to altered riparian habitats [38,39]. The obtained associations between bird assemblages and different land use gradients that can be equated with habitat disturbance, corroborate that particular bird assemblages can be considered as bioindicators of riparian integrity, and reflect the extent to which different land uses affect riparian bird communities. This was showed by the positive relation of the abundance of farmland, ground-foraging and granivore species with anthropogenic land uses (farmlands, urban areas, acacia stands) and the negative relation with more natural habitats (deciduous riparian forests, shrublands, pinewood and oakwood). Conversely, the abundance of invertivores and tree foragers were positively related with natural land uses and negatively related with altered uses, such as farmlands, urban areas or acacia stands.

The gradient of deciduous forest connected to shrubland over farmland, which represents open habitats and anthropogenic disturbance of riparian galleries, negatively and significantly affected bird assemblages all year round, with the exception of ground foragers during the wintering season. We noticed an absence of positive effects of this gradient on invertivores, known to be a good indicator of riparian integrity [38]. We believe that an absence of a positive relation with this group can be justified by invertivores in riparian corridors being affected by the presence of crops in the surrounding areas, since farmlands were frequently connected to riparian galleries in our study area. This is supported by Martin et al. (2006) [72], who stated that this habitat is sensitive to modification of adjacent landscapes. Also, in their study in a Mediterranean riparian gallery, Pereira et al. (2014) [25] conclude that the surroundings have an influence on bird assemblages during the breeding season. On the other hand, our study not only suggests that these effects occur in breeding populations but that they also affect wintering and migratory populations. The lack of any significantly positive relations can indicate that the vertical connectivity between riparian deciduous forests and shrublands of Erica spp., Ulex spp. and Cytisus striatus does not benefit the forest-related and invertivores bird assemblages.

The abundance of forest birds and omnivores was favoured by the woody associations of pine, oak and poplar tree stands during migration. During winter, this gradient was particularly important for the increase in abundance of invertivores and tree-foraging birds. In the breeding season, the gradient continued to influence the abundance of invertivores and was also important for ground foragers. The positive relation with understorey birds during the breeding season indicates that this forest association, which integrates some production stands of (mainly) exotic poplar trees, still contains a well-developed understorey layer. Usually, forest production management removes the understorey and shrublands of these stands [73]. When properly managed and without applying intensive control on the associated vegetation, native stands for production can provide good conditions for the establishment of native bird species [74].

Our results reflected the harmful effects of forest stands dominated by invasive species on bird communities in riparian corridors. The eucalyptus stands over urban area gradient was negatively related with the abundance of granivores during all seasons, ground foragers during the breeding season, unspecialised foragers and non-breeding birds in winter, forest birds in winter and farmland birds in the breeding season. The eucalyptus spreads and overtakes native forests due to its high growth capacity [75,76]. Additionally, these species bloom during winter, which can cause the displacement of food availability and ideal habitat conditions for birds when replacing native species [77]. A similar relation occurs with stands of Acacia spp. The increase in these invasive species was the gradient of highest importance in explaining the negative effects on the abundance of forest birds during the migration period and the breeding season. It also negatively impacted generalists and other habitat specialists and ground foragers in the breeding season. Although the abundance of two bird assemblages—farmland birds and ground foragers—was positively related with this gradient in the migration season, since the acacia stands were generally on the edge of farmland, we believe this association is a result of edge effects [78]. We also considered the positive effects on the abundance of farmland birds and ground foragers as an indicator of anthropogenic disturbance.

The process of ecological succession transforms these former croplands into low-shrubland areas [79], with tall herbaceous plant associations and sparse younger stages of pioneer woody species. The abandonment of farmlands present in riparian corridors increases the abundance of invertivores and tree foragers in winter and in the breeding season, respectively, and it negatively affects carnivores in the migration period and the abundance of granivores, ground foragers and farmland birds in the breeding season. Soil moisture and the typical mild winter climate of the region likely contribute to the abundance of invertebrates, an important trophic resource for invertivore and omnivore species [68]. In these habitats, we observed high densities of small arthropods, mostly arachnids, in the vegetation during the fall and winter campaigns (personal observations). Some species like Cetti’s warbler and the dunnock (Prunella modularis) were frequently detected in these areas.

5. Conclusions

This study adds to our understanding of the dynamics of riparian bird communities in northwestern Iberian Peninsula and supports the use of functional bird assemblages to assess the effects of disturbed landscapes. It also corroborates other studies on riparian corridors, suggesting that these sensitive habitats are fundamental to maintaining a diversified and rich avian community in human-dominated landscapes [80]. This study contributes to filling the gap in our knowledge of the Atlantic riparian bird communities in Portugal, as the only studies we are aware of have been conducted in Mediterranean riparian corridors [25,29]. Additionally, we provide information on the effect of invasive woody species, namely eucalyptus and acacias, and the possible effects of these land uses. To our knowledge, only one study has addressed the impact of invasive wattle acacia stands on local bird communities in the Iberian Peninsula [81], making our findings relevant and a starting point to future research.

For further studies, we suggest the application of environmental data that reflect the structural and ecological features of riparian corridors, such as riparian width, the percentage of invasive plant species and coverage of canopy, understorey and herbaceous vegetation. We also consider that the determination of possible edge effects via proximal land uses in riparian corridors would be of extreme importance for riparian landscape ecology studies and conservation.

We also emphasise the priority of conservation of riparian corridors and their adjacent natural habitats through protection of these ecosystems or through rehabilitation of disturbed riparian areas. For this purpose, we reinforce the possibility raised by Larsen et al. (2010) [38] of integrating riparian birds’ assemblages as indicators under the Water Framework Directive (WFD), since this programme uses biological measures to assess riverine ecosystems’ ecological status [82].