Density variation in “rare” breeding birds in native forests and urban parks

Density variation in “rare” breeding birds in native forests and urban parks. — O. Dubovyk, H. Kuzyo, A. Bokotey. — Biodiversity protection and conservation of rare species are typically the main goals of protected areas. Protected areas are commonly created within native ecosystems where anthropogenic pressure is low. Meanwhile, a growing body of literature has focused on the effectiveness of protected areas to provide important habitats for rare taxa. To highlight this issue, we examined avian species, the most diverse vertebrate class in Ukraine. Forest ecosystems vary relative to conservation status and native integrity. With this in mind, we considered Lviv, Ukraine and outskirts. To determine conservation status, we used regulatory instruments such as the National Red Book of Ukraine, Resolution 6 of the Bern Convention on the Conservation of European Wildlife and Natural Habitats and appendices, the Bonn Convention on the Conservation of Migratory Species of Wild Animals and appendices, and Species of European Conservation Concern (SPEC) status to define rare species. The weighted density of SPEC species and those listed in the Bern Convention and Bonn Convention vary considerably and are higher in forests where native integrity has been lost to various degrees. The density of rare taxa within forests exhibiting high biodiversity loss (e.g. urban parks) was starkly evident. Degree of urbanization, as a function of distance from Lviv, was an insignificant predictor of a species listed in Resolution 6 of the Bern Convention or Red Book of Ukraine. We found that weighted density of species with mean species statuses weight were lower in more native forests than parks. Given that weighted densities were highly correlated with general density, we attribute this finding to the luxury effect. That is, density of birds is higher in more urbanized areas of a particular habitat type (e.g. forests and parks). Our conclusion supports previous findings that an increase in overall bird density is common among European cities. To a lesser extent, these findings suggest ineffective nature conservation management of forests in the Lviv region of Ukraine and imply the lack of interest in urban parks as habitat for rare taxa.


Introduction
Regulations on the consumption of natural resources have a long history.A restriction on natural resource use began more than 2000 years ago in ancient Asia and more recently in Medieval Europe for hunting by monarchs and feudals.Since the 19th century, this protection of resources has progressed to what we now know as nature conservation areas, including such modern examples as Yosemite or Yellowstone National Parks in the United States (Phillips 2004).
The impetus to establish conservation areas are varied.On the one hand, protected areas (PAs), any areas protected by a national law such as national parks or nature reserves, are established to preserve rare species.At times, the protection of species may be through umbrella species where, for example, protection of a predator may result in the protection of its prey (Simberloff 1998).On the other hand, PAs are an important instrument for biodiversity conservation, which is needed against р-ISSN 2617-6157 е-ISSN 2617-6165 GEO&BIO • 2020 • том 19 the backdrop of species extinctions and potential lack of biological resources (Lopoukhine 2008).Ukrainian nature conservation law gives wide and opaque latitude when defining the rationale for the creation of PAs: to save the natural diversity of landscapes and gene pool of animals and plants, to support ecological balance, and to conduct environmental monitoring.Oddly, neither biodiversity nor rare species protection are mentioned, although they can be alluded to when attempting to save gene pools.
The effectiveness of PAs and their functioning and management is questionable (Arcese & Sinclair 1997).For instance, there is poor overlap between potentially suitable habitat of globally threatened birds of mainland Africa and PAs (Beresford et al. 2011).Likewise, there is a mismatch between suitable habitat for plants and the occurrence of PAs (Burgess et al. 2005).Even more striking are the numerous examples where rare animal taxa do not even use PAs (Rodrigues et al. 2004).Ironically, there are also areas without any formal status as a nature conservation site that serve a similar function of PAs (Bhagwat et al. 2005).Despite these telling facts, the evidence does not suggest an ineffectiveness of modern PAs, but rather, a message that the management of PAs are in need of improvement (Rodrigues et al. 2004;Burgess et al. 2002).
Therefore, a question arises: do PAs perform their functions effectively in protection of rare species and especially in providing for future long-term survival?To answer this question, we focus on birds, the most diverse group of vertebrates in Ukraine and a taxonomic group that serves as an indicator and which are relatively convenient to study (Gorban & Tsaryk 2012).The comparative study of the density of rare bird species can help to elucidate general patterns of habitat preference.
The majority of studies use a more traditional approach to define both rare species and PAs.There is, however, a lack of research on the distribution of rare taxa within urban gradients of native integrity.This alternative approach appropriately captures this relationship between native integrity and the effectiveness of PAs.As PAs are commonly created in ecosystems of a high degree of native integrity, the use of native integrity gradient or the degree of urbanization gradient is more convenient than "protected areas' regime acerbity" gradient.
Therefore, we examined forest communities of breeding birds that exhibit the highest biodiversity in western Ukraine.The comparative element lays in the way of studying forest ecosystems in "urbanization -native integrity" gradient (or urban gradient sensu Blair (Blair 1996), but despite the fact that urban gradient implies "from more native to more urbanized", we used "from more urbanized to more native" approach).The aim of this work was to reveal the preferences of rare taxa within this gradient by a comparison of density of rare species in forested park ecosystems within city limits relative to native forests in nature reserves outside the city.

Study area
The study area included Lviv city within its ecological (i.e. at the edge of the urban matrix where it meets the non-urban interface) and administrative boundaries, including suburbs and adjacent territories (Fig. 1).
Material for this work was collected during the breeding period (Apr -Jun) in 2018 and 2019.Twelve parks and cemeteries were chosen in geographically representative areas of Lviv.We also included two forests in the suburbs and two PAs distant from the city.Linear transects of 500 m in length and 100 m in width were established within each plot.Number of transects per plot depended on the area of the plot: 1 to 10 transects per plot (3.7 transects per plot in average), N = 59 transects.
We conducted bird counts during the breeding season (15 Apr -20 Jun) every ~30 days three times on each transect.A typical survey was conducted within 5 hrs after sunrise.Species were identified by sight and sound and all individuals within 50 m of the transect were recorded.Transects were traversed at a mean rate of ~1 kph.We assigned a degree of urbanization value to every transect (Fig. 1, Table 1), irrespective of whether the transect was within a PA, because PA statuses are irrelevant in some cases.For example, there are two regional landscape parks among model plots: Znesinya is located near the centre and is a popular place for recreation with developed infrastructure.The other is Stilske Gorbogirya, which is located 10-30 km south of the city and consists of mature broadleaf forests used for forestry.
Plots were grouped as follows: • Parks in the centre of the city (11 transects in Ivan Franko Park, Vysokyi Zamok Park and Znesinnia Regional Landscape Park partly, Zalizna Voda Park, Lychakiv Cemetery, Stryisky Park); • Parks in the periphery of the city (17 transects in Znesinnia Regional Landscape Park partly, John Paul II Park, Pogulianka Park, Snopkivsky Park, Holoskivske Cemetery, Sknyliv Park, Yaniv old Cemetery, Lychakiv park, Bilogorshya Forest partly); • Forests in suburbs (15 transects in Bilogorshya Forest partly, Briukhovychi forest, Vynnyky forest); • Forests distant from the city (16 transects in Stilske Gorbogirya Regional Landscape Park and Roztochya Nature Reserve).Although this classification is approximate, it facilitated data analysis because the number of transects was relatively equal among groups.We call the classes listed above the zones of degree of urbanization here and later.It is notable that some plots are listed in different zones related to habitat heterogeneity along linear transects within the plot (Table 1).

Classification of rare species
Legal and regulatory instruments were analysed to assign rarity designations among species.We did not assume that the formal status of a species implied its ecological rarity (Gaston 1994).We did not take into account IUCN status when calculating the mean weight of each status because all taxa had the same status (Least Concern) except the European turtle dove Streptopelia turtur (L.).
We used Species of European Conservation Concern (SPEC) categories to assess the rarity of species in Europe (BirdLife International 2017): Non-SPEC (species whose global population is not concentrated in Europe, and whose European population status is currently considered to be Secure); SPEC 3 (species whose global population is not concentrated in Europe, but which is classified as Regionally Extinct, Critically Endangered, Endangered, Vulnerable, Near Threatened, Declining, Depleted or Rare at European level); SPEC 2 (species whose global population is concentrated in Europe, and which is classified as Regionally Extinct, Critically Endangered, Endangered, Vulnerable, Near Threatened, Declining, Depleted or Rare at European level); and SPEC 1 (European species of global conservation concern, i.e. classified as Critically Endangered, Endangered, Vulnerable or Near Threatened at global level).
We used the Bern Convention on the Conservation of European Wildlife and Natural Habitats (Bern Convention, BC) and Bonn Convention on the Conservation of Migratory Species of Wild Animals (Bonn Convention, CMS) among international documents that are implemented in Ukraine.We took into account the presence of species in Appendix II (Strictly protected fauna species), Appendix III (Protected fauna species) and Resolution 6 (listing the species requiring specific habitat conservation measures) of BC and presence of species in Appendix I (Threatened migratory species) and Appendix II (Migratory species requiring international cooperation) of CMS.
The statuses of the third edition of the National Red Book of Ukraine (RB) were used to determine status relative to national laws.
We used an approach similar to that used by Zagorodniuk (2008): the categories of each conservation status ranged from 0 to 1. Thus, RB status was as follows: 1.00 = Vulnerable, 0.75 = Rare, 0.50 = Undefined, and 0.25 = Not Evaluated.Categories of SPEC were as follows: 1.00 = SPEC 1, 0.67 = SPEC 2, 0.33 = SPEC 3, and 0.00 = Non-SPEC; presence in Appendix II of BC = 1.0,Appendix III = 0.5.Presence on the list of Resolution 6 of BC (R6) was considered separately and given a weight of 1.0.The presence in one of two appendices of CMS was given a weight of 0.5, but if present in both a weight of 1.0 was used.
Additionally, we used the mean of weights (MW) as a measure of conservation importance of species, despite the fact that this system of evaluation was used more as a matter of convenience.Species status, distribution among plots and mean weights are listed in Table 2.

Statistical analysis
Statistics were calculated in R 4.0.1.(R Core Team 2020) using the RStudio integrated development environment (1.3.959) and libraries xtable, usedist, rcompanion, ggpubr, gvlma, mblm.The map was designed in QGIS (Open Source Geospatial Foundation Project) using satellite imagery of Bing and Google Maps with postprocessing in FastStone Image Viewer (FastStone Soft).
Data were transformed for analyses to reduce dimensionality into eigenvectors.The status weights were used to reduce the transect densities for each species to weighted sums of densities.
The raw data were presented as a matrix of density values (pairs per transect, equivalent to a belt transect area of 500 m × 100 m = 5 ha) of species by transect.The density of a species on a transect was presumed to be equal to the maximum number of pairs observed during three consecutive counts.Before the data were transformed, they were presented as a matrix of density values d (pairs/5 ha) of the species s on the transect t, or d s, t .In addition, each species had the defined status weight (one of six conservation statuses CS) w CS .To calculate the weighted densities of species we transformed the initial matrix into six matrices with values of d s, t •w CS .This action allowed us to summarise weighted densities for each transect taken as Σd s, t •w CS = d 1, t •w CS + d 2, t •w CS + … + d S, t •w CS for S species registered on the transect.Thus, we received six values per transect that corresponded to each conservation status and were used in further statistical analysis.
The census data collected in previous studies (from 2012 to present) was also used to assess the species diversity on plots.These data were not used in statistical analysis, but instead were used in Table 2.This information can be essential because transect method implies ignoring all birds outside the transect, therefore we get relatively precise density data, but species list could be incomplete.
We used Shapiro-Wilk test to check the normality of raw data -vectors of weighted values.Also, to check a suitability of parametrical methods, we used this test to check the normality of residuals of ANOVA and linear model.Zone of degree of urbanization was used in ANOVA as a predictor.Although, a simple linear model needs a continuous predictor, and we used the natural logarithm of distance between the transect and city centre (49°50'22.8"N,24°1'46.6"E) in this role.Linear models were also checked with global validation of linear model assumptions (Peña & Slate 2006) using gvlma in R. All of the above signified that parametrical methods could be used in limited cases in our data (Table 3 & 4).
Thereby, we used non-parametric analogous methods: Kruskal-Wallis test was used instead of ANOVA and Kendal-Theil Sen Siegel nonparametric linear regression instead of linear model.

Results
Primarily, the appropriateness of comparison of the logarithm of the distance from the centre with a zone of degree of urbanization has to be verified.Those logarithms are not distributed normally (W = 0.906, p < 0.001), so we checked this with Kruskal-Wallis test: KW χ 2 = 51.964,p < 0.001.Therefo re, we could use the logarithm of the distance from the city to assess the trend of weighted densities' changes throughout the urban gradient.
As it was mentioned above, not all raw vectors of weighted densities had a normal distribution neither residuals of parametrical models did.Thus, parametrical models are not acceptable for analysis of variations of weighted densities with weights of RB, R6, SPEC statuses used (Table 3).
Zone of degree of urbanization was not a good predictor of Red Book weighted density status in Kruskal-Wallis test (Table 5).Given that most "red-book" species in our data lacked any estimates of density (only 5 species' densities are known precisely: the stock dove Columba oenas L. and the Eurasian green woodpecker Picus viridis L. are Vulnerable, the white-backed woodpecker Dendrocopos leucotos (Bechstein) is Rare, the Ural owl Strix uralensis Pallas is Undefined, the common firecrest Regulus ignicapilla (Temminck) is Not Evaluated), we assumed that analysis of densities of those species would be inappropriate.
Table 5. Results of Kruskal-Wallis test with a zone of degree of urbanization as a predictor and results of Kendal-Theil Sen Siegel nonparametric linear regression with a logarithm of distance from the centre as a predictor; weighted densities were used as a dependent variable in both models Таблиця 5. Результати тесту Крускала-Валліса із зонами ступеню урбанізованості в якості предиктора та результати непараметричної лінійної регресії із логарифмованою дистанцією від центру міста в якості предиктора; зважені щільності були використані в якості залежної змінної в обох моделях The sums of weighted densities of species listed in appendices of the Bern Convention were related significantly to zones of degree of urbanization.Nonparametric linear regression supported the trend of this relation (see Fig. 2, b): the closer to the city centre transect is, the higher weighted density is observed.
Some species listed in Resolution 6 of Bern Convention preferred distant plots from the city (such as the white-backed woodpecker, the red-breasted flycatcher Ficedula parva (Bechstein), and the Ural owl) and others were distributed evenly throughout all plots (the collared flycatcher Ficedula albicollis (Temminck), the middle spotted woodpecker Leiopicus medius (L.), and the gray-faced woodpecker Picus canus Gmelin).In the latter case, the even distribution can compensate for the variance in weighted densities, which makes this analysis uninformative: it concerns the collared flycatcher especially, as this species is frequent in forests of western Ukraine (Hnatyna 2017) (row "R6-FA" in Table 5 means weighted densities with the collared flycatcher density ignored).In both cases, the significance of variation of weighted densities of R6 species throughout the urban gradient is unclear.When the collared flycatcher densities are not ignored (row "R6"), variation of weighted densities in different zones of degree of urbanization can be seen without any visible trend (Fig. 2, c).Pairwise comparisons using Wilcoxon rank test (with Benjamini & Hochberg p-value adjustment method) showed that weighted densities are significantly different in zones of degree of urbanization of 2 and 3 (p = 0.001), 2 and 4 (p = 0.034), 1 and 2 (p = 0.034).
The negative coefficient of a nonparametric linear model with CMS statuses weights implied that the abundance of species protected by CMS was lower on transects in more native forests (Fig. 2, d).Weighted density of SPEC species was significantly predicted by zones of degree of urbanization with a significant decline in more native forests (Fig. 2, e).Zone of degree of urbanization was a significant predictor of SPEC weighted densities between zones 1 and 4 (p = 0.004) and 2 and 4 (p = 0.004) (pairwise comparisons using Wilcoxon rank test).
The mean weight of conservation status can be described as an integral conservation value of a species.However, it gives the same value to all the conservation statuses we used, so it makes our formal approach to assess the rarity of species even more formal.As it was in previous cases, weighted densities significantly vary on different zones of degree of urbanization and the trend of this variance is to decrease in more native forests.
When the overall density of breeding birds is analysed (rows "OD" in Tables 3-5), it can be seen that all of the significant results above are the same as the trend of birds' overall density in urban gradient: the overall population of birds, irrespective of their taxa and their conservation status, is more dense in parks of the city than in native forests.
Correlation of weighted densities of those statuses that returned significant results with an overall density vector is also significant.Thus, Pearson correlation between BC weighted densities and overall densities is R = 0.919 (p < 0.001), between CMS weighted densities and overall densities is R = 0.885 (p < 0.001), between SPEC weighted densities and overall densities is R = 0.927 (p < 0.001).

Discussion
Our conclusion, in general, is that the density of rare species is lower outside the city, in native forests that belong to PAs.However, the density of breeding birds of all species was higher in urban parks than in native forest.
Such findings might be attributed to a difference in track lengths if we had used routes of different lengths.However, the study was designed to avoid such problems, so all of the study units used linear transects with equivalent areas and lengths.We assume that densities of breeding forest birds certainly depend on transect position, so that a higher density may be expected in urban parks more so than in native forest outside the city.
It is notable that the significant results returned by statistical models concerned those statuses that are commonly assigned to species.Weighted densities with weights of BC, CMS and SPEC statuses, whose negative regression coefficients were significant, correlated significantly with density of all species, which is obvious because they were derived from the last one.When we analysed weighted densities by the mean weight of status, there were not so many species that did not have any conservation status (see Table 2), so it is obvious that a big part of variance remains after data transformation and models return significant results.
It is assumed that species diversity and quantitative density can be higher in cities in the works of different authors.The species diversity cannot be discussed from the positions of our work to compare our results with others, additionally, more habitat types except forest ecosystems should be analyzed to make objective conclusions (Blair 1996).With regard to density, in general, it is thought that moderately built-up areas contain bird communities with higher densities (Tratalos et al. 2007).This fact can be explained by the additional food resources, local microclimate associated with urban habitats and invasion patterns of urbanized species (Møller et al. 2012).Higher densities of breeding birds have been previously documented in more urbanized areas.For example, bird abundance may increase across an urban gradient with a bimodal distribution and the highest abundance of invasive species in the most urbanized areas and the highest abundance of indigenous species in residential areas with high vegetation (Clergeau et al. 1998).An increase in overall bird density is a common phenomenon in central and north-western European cities (Tomiałojć 1998).Avifauna of eastern European cities, such as Lviv, shares this feature as evidenced by our findings, although the timing (when exactly this fact became true) is under question because of the lack of studies in this field.
The above mentioned touches the topic of the "luxury effect", but researches concerning this ecological effect are usually applied to the whole urban and non-urban habitat with a diversity of habitat types (Chamberlain et al. 2019).In contrast, we have studied only forest-like habitats, which explains the linearity of density changes in urban gradient.

Conclusions
Within the region of Lviv and its outskirts, the densities of forest breeding birds listed in the appendices of the Bern Convention on the Conservation of European Wildlife and Natural Habitats and the Bonn Convention on the Conservation of Migratory Species of Wild Animals, as well as of species with Species of European Conservation Concern status, exhibit a strong, positive relationship with a degree of urbanization of a forest.We were unable to assess species with statuses in the National Red Book of Ukraine and Resolution 6 of Bern Convention because of insufficient sample sizes for quantitative assessments.
Our approach, taking into account several conservation statuses of species, shows that species who are formally designated as rare species (i.e.species with many conservation statuses) are at higher densities in urban parks than in native forests.
We are not implying that protected areas serve no value, but our findings suggest that conservation management of forests within the Lviv region is ineffective for the conservation of rare forest breeding birds.More confidently, we assume that our results point at the general density patterns of breeding birds within urban gradients, at least for single forest-like habitat types.In terms of nature conservation, urban parks may play an essential role as valuable nature conservation sites, especially for the conservation of rare species.From the perspective of urban ecology, bird community density patterns are being affected by anthropogenic pressure, leading to a corresponding change in the density of rare species.