Birds are excellent biodiversity indicators because their population trends are affected by many ecological factors, such as the niche breadth (SEOANE, LAIOLO, & OBESO, 2017) and the climatic niche (JIGUET ET AL., 2010). They are very sensitive to environmental changes (BLAIR, 1999; FULLER ET AL., 1995; RUSSO, 2007), although sensitivity differs among species (MCGILL ET AL., 2015; SIRAMI, ET AL., 2008). But also birds constitute a conspicuous and cosmopolite animal group generally easy to sample (ESCANDELL, 2013). Due to these reasons, birds are commonly used as model species in long-term monitoring studies of diversity changes (ALLEN ET AL., 2019; PARODY ET AL., 2001; SIRAMI ET AL., 2008) or population trends (ŞEKERCIOĞLU ET AL., 2019; TRABA & MORALES, 2019).

Human-induced environmental changes are impacting biodiversity (BOWLER ET AL., 2020; JETZ ET AL., 2007; KERR & CURRIE, 1995; SHIVANNA, 2020), such as bird species richness and abundance (DA SILVEIRA ET AL., 2016; FONDERFLICK ET AL., 2013; REDLICH ET AL., 2018). For example, many western humanized areas have seen a gradual rural abandonment of traditional agriculture throughout the 20^th century (BEILIN ET AL., 2014; CORBELLE-RICO ET AL., 2012; MACDONALD ET AL., 2000; REGOS ET AL., 2016), while others have experienced the process of industrial agricultural intensification (DONALD ET AL., 2001; GAO ET AL., 2019). As a consequence, in both cases, biodiversity levels might decrease as has already been observed in plant communities in different regions of Europe (BUHK ET AL., 2017; FLOHRE ET AL., 2011; GUITIÁN ET AL., 2012; SIRAMI ET AL., 2007) but also in bird communities (BATÁRY ET AL., 2018; DONALD ET AL., 2001; FLOHRE ET AL., 2011; FULLER ET AL., 2007; GUERRERO ET AL., 2012; SALAVERRI ET AL., 2019; SIRAMI ET AL., 2008).

Over the last 50 years, rural Europe, and within it, the NW of the Iberian Peninsula, have undergone a strong process of rural abandonment. Within the natural region of northwestern Spain, O Courel (42° 34′ 59.99″ N, 7° 10′ 0.01″ W; 21,020 ha), the number of inhabitants was reduced (-60%) and with it, the livestock (-80%) and number of farmlands (-36%) decreased as well (LÓPEZ-BAO ET AL., 2015; MUNILLA ET AL., 2008). All these changes have modified vegetation patterns favoring woodland areas (LÓPEZ-BAO ET AL., 2015; MUNILLA ET AL., 2008), with an associated loss of bird diversity (SALAVERRI ET AL., 2019) which may be linked to a landscape homogenization (BUHK ET AL., 2017; SUÁREZ-SEOANE ET AL., 2002).

However, within O Courel region, areas which have not been cultivated remained stable, thus a vast majority of the region has not locally changed (VALERO ET AL., 2019). We do not know whether and how bird richness and abundance may have changed in the well-preserved areas of this region after a 37-year period. Analyzing changes in the richness and abundance of bird communities after several decades in well preserved plots in NW Iberia would help us understanding bird diversity trends at undisturbed sites, i.e. which have not undergone abandonment or industrialization.

We investigated changes in bird diversity and species trends after a long-term period in different habitats which have remained well preserved within O Courel region. Here, we assessed changes in overall bird species richness and abundance within five sites (two chestnut woodlands, a montane forest, a holm oak forest and a heathland) after 37 years, using data obtained from bird surveys performed in the same locations with the same methodology in July 1977 and July 2014. Bird populations might show different trends over time, and this could be related to their specific traits (ALLEN ET AL., 2019; FONDERFLICK ET AL., 2013; LEHIKOINEN ET AL., 2019; MORELLI, ET AL., 2020; SIRAMI ET AL., 2008). Thus, we additionally analyzed whether species phenotypes are related to the abundance trends of the 15 most abundant species overall in this area. We hypothesized that (i) the overall number of species and bird abundance varied after a 37-year period, (ii) and these changes may differ among sites, and (iii) changes in species abundances may be explained by bird phenotypic traits.

Overall, 392 individuals of 31 bird species were recorded.

The number of bird species varied among sites but did not change overall from 1977 to 2014 (Table I, Fig. S1a). The heathland was the site with the lowest specific richness (6 species in total for both years). Similarly, the overall abundance of birds (sum of individuals per site and year corrected by the number of surveys) has not changed after the 37-year period although it differed among sites (Table I, Fig. S1b), being the heathland the site with lower bird abundance (8 birds in total for both years).

Table I

Response variable	Fixed effects	Wald Chi-square	d.f.	P-value
Richness (no. of species)	Year	1.434	1	0.231
	Site	19.242	4	0.001
Abundance (no. of individuals)	Year	0.802	1	0.370
	Site	19.156	4	0.001

Among the 15 more abundant species (i.e. more than 10 individuals observed in both years), only eight species showed a change in their abundance over time and 13 species differed in their abundance among sites (Table II). In particular, the abundance of Cyanistes caeruleus, Garrulus glandarius, Parus major, Prunella modularis and Sylvia atricapilla increased between 1977-2014, and Erithacus rubecula, Phylloscopus ibericus, and Troglodytes troglodytes decreased in their abundances after the same 37-year period (Table II, Fig. 1). Among the seven species that did not differ in abundance between years: Fringilla coelebs, Periparus ater, Regulus ignicapilla, Sylvia undata and Turdus merula showed differences in the number of individuals detected per species among sites, as did the 8 species that varied in abundance over time (Table II, Fig. 1).

Table II

Response variable		Species	Fixed effects	Wald Chi-square	d.f.	P-value	Change
Abundance (no. of individuals)		C. caeruleus	Year	5.330	1	0.021	+
			Site	26.408	4	<0.001
		E. rubecula	Year	5.358	1	0.021	-
			Site	225.432	4	<0.001
		F. coelebs	Year	0.357	1	0.550	ns
			Site	14.988	4	0.005
		G. glandarius	Year	24.323	1	<0.001	+
			Site	223.990	4	<0.001
		P. ibericus	Year	6.265	1	0.012	-
			Site	10.000	4	0.040
		P. major	Year	18.045	1	<0.001	+
			Site	13.068	4	0.011
		P. modularis	Year	10.808	1	0.001	+
			Site	31.557	4	<0.001
		P. ater	Year	0.908	1	0.341	ns
			Site	14.571	4	0.006
		R. ignicapilla	Year	3.623	1	0.057	ns
			Site	24.371	4	<0.001
		S. atricapilla	Year	6.552	1	0.010	+
			Site	38.620	4	<0.001
		S. europaea	Year	2.071	1	0.150	ns
			Site	9.142	4	0.058
		S. serinus	Year	0.000	1	0.986	ns
			Site	8.349	4	0.080
		S. undata	Year	2.500	1	0.114	ns
			Site	20.408	4	<0.001
		T. merula	Year	0.873	1	0.350	ns
			Site	11.695	4	0.020
		T. troglodytes	Year	12.59	1	<0.001	-
			Site	13.59	4	0.009

Figure 1

With regard to bird traits analysis, we did not find any relationship between the phenotypes of bird species (measured by PCA from the phenotypic space using 27 traits, Table SII in Supplementary section) and the specific change in species abundance over 37 years for the 15 more abundant species (Table SIII, Fig. 2).

Figure 2

The overall number of bird species and individuals did not change after 37 years, in well-preserved areas from NW Spain (O Courel mountains) but we found specific changes in abundance of 8 of 15 most abundant species. These species specific trends may depend on their characteristics, their preferences for a certain land cover, their migration patterns or their biogeographical origin (SUÁREZ-SEOANE ET AL., 2002). For example, species traits may determine their level of success in particular habitats, where certain level of specialization is required for breeding and surviving (ALLEN ET AL., 2019; COETZEE & CHOWN, 2016; LEHIKOINEN ET AL., 2019). However, we did not find a relationship between species traits and abundance change in 15 most abundant bird species after 37 years, maybe due to the lack of sufficient phenotypic variation among these 15 species compared here.

We found differences in species richness and overall bird abundance among sites, being the heathland the site with the lowest number of species and abundance compared to the other sites. These differences observed among sites may be linked to differences in floristic composition and vegetation structure (DÍAZ, 2006; GUITIÁN ET AL., 2004; ROSENVALD ET AL., 2011), as well as, land cover (DA SILVEIRA ET AL., 2016). Thus, the higher number of species observed corresponds to mature deciduous forests (i.e. with a more diverse plant community and complex vegetation structure): Moreda’s (20 species in total) and Parada’s (19 species) chestnut woodlands, the montane forest (17 species overall), and the holm oak forest (14 species in total), but the lowest species numbers was found in the heathland (6 species overall), the site with lower plant diversity and more simple vegetation structure. However, bird species numbers found in our mature deciduous forests were rather lower than those found in other NW Iberian forests (30-37 species; BAS ET AL., 2018; CALVIÑO-CANCELA, 2013; GODED ET AL., 2019), which may be explained by altitudinal differences with regard to our study sites, that have already accounted for a decrease in bird species at higher elevations (SALAVERRI ET AL., 2019), specially forest-dwelling species (TELLERÍA, 2019), but also by the landscape structure, in particular when considering the surrounding areas (FONDERFLICK ET AL., 2013). Thus, marked habitat differences among sites may explain the reduced number of bird species and abundance found in the heathland.

Moreover, the greater numbers of birds (as the sum of individuals per site and year corrected by the number of surveys) were found in the Moreda’s chestnut woodland (28 birds in total for both years), montane forest (27 birds in total), and the holm oak forest (23 birds in total), and these numbers are rather lower than those observed in the already mentioned NW Iberian forests (60-140 total birds; BAS ET AL., 2018; CALVIÑO-CANCELA, 2013; GODED ET AL., 2019). If we compare bird abundance among study sites, our results may suggest that mature deciduous forests could be acting as bird diversity reservoirs due to a possible landscape homogenization of the surrounding plots (BRANDT ET AL., 2013), as a consequence of the ongoing farmland abandonment process (HERZON ET AL., 2014; SALAVERRI ET AL., 2019; VERHULST ET AL., 2004), but also the increase of woodland areas in this region (LÓPEZ-BAO ET AL., 2015; MUNILLA ET AL., 2008). These results may be supported by the fact that O Courel still maintains well preserved areas given the landscape heterogeneity provided by a mosaic of large patches, mainly dominated by leaf deciduous forests and scrublands (VALERO ET AL., 2019).

Therefore, the bird richness and abundance remained stable after 37 years at the study sites in O Courel Mountains, despite the specific changes in species trends of 8 most abundant species found within our study area. Although we were not able to analyze the effect of the year*site interaction (given the level of our analysis), it seems that these eight species specific trends might differ when we separate the change in species abundance by site (Fig. 1). One possible explanation to the specific changes in species abundance is the habitat preference of these species (GARCÍA ET AL., 2014). The forest-dwelling species (Cyanistes caeruleus, Garrulus glandarius, and Parus major) showed a generalized increase, which is consistent with the increased area occupied by forest in this well preserved region over the study period (LÓPEZ-BAO ET AL., 2015; MUNILLA ET AL., 2008). In contrast, bird species highly associated with the underbrush (such as Prunella modularis, Sylvia atricapilla, and Troglodytes troglodytes) have increased in the Moreda’s chestnut forest, the montane forest and the holm oak forest, which could indicate that the shrub substrate has increased in these deciduous forests. Surprisingly, Phylloscopus ibericus consistently decreased in all the study sites, which might be explained by the relevant population fluctuations of trans-Saharan migratory species observed (PAYEVSKY, 2006). Lastly, the general declined observed in Erithacus rubecula and Troglodytes troglodytes (excepting at the Moredas’s chestnut woodland), but also the site differences observed in other species with no changes over time (Fringilla coelebs, Periparus ater, Regulus ignicapilla, Sylvia undata and Turdus merula) are difficult to interpret, partly due to site-specific factors that could be acting at each site and promoting different dynamics of species population trends.

Finally, the absence of a significant effect of species phenotypic traits on the change in abundance experienced by the 15 most abundant species can be explained by the fact that these 15 species did not differ enough from each other (in the phenotypical space studied with 27 avian traits, which were mainly body size and breeding strategy) to detect a relationship of temporal patterns in species abundance and bird phenotypic traits. This result might suggest that different bird traits to that included in our species phenotypes analysis, could explain specific changes observed in our 15 species over time, such as the habitat preference or breeding habitat specialization (MORELLI ET AL., 2020), but also migration distance (JIGUET ET AL., 2010; SIRAMI ET AL., 2008) or even biogeographical origin (SUÁREZ-SEOANE ET AL., 2002).

To conclude, our study shows no changes in bird richness and overall bird abundance after 37 years in a well conserved mountain area of NW Iberia (O Courel) and finds changes in species specific trends (increasing trends in 5 from the 15 more abundant species, mainly forest-dwelling species, and decreasing trends in 3 species) which are unrelated to the species phenotypic traits. Our results indicate that bird communities in this well conserved region are maintaining their levels of overall diversity, despite possible site-specific shifts in species abundances. Further investigations in well preserved areas long-term linked to human presence will be relevant to understand the rate of change in biodiversity at local scales in the context of a globalized world.