Figure
1.
Citywide survey map of Xalapa showing the distribution of sampled point counts
| Citation: |
Jörg HOFFMANN, Udo WITTCHEN, Ulrich STACHOW, Gert BERGER. 2013: Identification of habitat requirements of farmland birds based on a hierarchical structured monitoring scheme. Avian Research, 4(4): 265-280. DOI: 10.5122/cbirds.2013.0026
|
Agricultural landscapes are essential for the conservation of biodiversity. Nevertheless, a negative trend continues to be observed in many rural areas for the most prominent indicator species group, the farmland birds. However, clear cause-effect relationships are rarely reported and sometimes difficult to deduce, especially from monitoring data which are based only on the detection of species and counts of the numbers of individuals. Because the identification of habitat preferences is a precondition for farmland bird biodiversity conservation efforts, a monitoring scheme for the simultaneous collection and analysis of bird and land use data was developed and tested. In order to assign the occurrence of bird species to land characteristics at various spatial scales and different land use and crop types, we applied a hierarchical structured sampling design. The spatial scales were 'agri-cultural landscape', 'agricultural landscape types', 'field crops and other habitats' and 'vegetation structures'. These scales were integrated with a novel concept, the 'habitat matrix' (HM). This method was applied to farmland breeding bird abundances on 29 plots, each 1 km2 in size, by the use of the territory mapping method. The same plots were enlarged by a 100 m buffer and the sizes and location of habitats documented. Vegetation height, coverage and density were also recorded for all crop fields in the study area. We propose that this monitoring method facilitates the identification of scale dependent relationships between farmland bird habitat characteristics and bird abundance. This is demonstrated by the farmland bird species Corn Bunting (Emberiza calandra), Skylark (Alauda arvensis), and Whinchat (Saxicola rubetra). The breeding territories of these species reveal large differences within the various spatial scales 'agricultural landscape', 'agricultural landscape types' and 'field crops'. Throughout the breeding season the abundances varied, dependent on the field crop and the development of vegetation structures (height, coverage, and density). HM-analysis led to the identification of specific habitat configurations preferred by individual bird species within the agricultural landscape. These findings indicate that the methodology has the potential to design monitoring schemes for the identification of cause-and-effects of landscape configuration, land use and land use changes on the habitat suitability and abundance of farmland birds.
Urbanization is one of the most important forces of global transformation (Picket et al. 2001; Grimm et al. 2008). Urban growth is rapid and pervasive (Paul and Meyer 2001), implying the modification, and even replacement, of preexisting conditions (Eldredge and Horenstein 2014). Besides the unmeasurable environmental impacts of urban metabolism at broad scales (Kennedy et al. 2011), cities represent systems with the widest array of pollutants, including solid waste, chemical pollution of air, water and soil, visual contamination, electromagnetic concentrations, and anthropogenic noise (Maldonado 2009). Given the environmental pressures that urbanization poses on biodiversity, it has been identified as one of the main causes of species endangerment and local extinction (Czech et al. 2000; McKinney 2002).
Although urbanization represents a semi-permeable ecological barrier for species from regional pools to colonize (MacGregor-Fors 2010), the set of species able to cope with the implied hazards and that survive on the available resources, among other factors, have shown to adjust, and even evolve, with urbanization (Johnson and Munshi-South 2017). Yet, evidence indicates that many of the urban stimuli are deleterious even for urban wildlife (Beaugeard et al. 2019). Among these stimuli, anthropogenic noise and artificial light at night have been considered crucial in understanding the response of birds to urbanization (Fröhlich and Ciach 2019).
Anthropogenic noise is regarded as a pollutant that can drive the behavior of species that rely on acoustic communication (Parris et al. 2009; Barber et al. 2010; Goodwin and Shriver 2011; Nemeth et al. 2013; Luther et al. 2016). Numerous animal species rely on acoustic signaling to perform some of the most elementary and complex processes, such as: (i) sexual signaling, (ii) territorial defenses, (iii) predator deterrence and/or avoidance, (iv) parental strategies, (v) foraging, and (vi) parent–offspring communication (Sanborn 2008; Dudzinski et al. 2009; Parris et al. 2009; Jacot et al. 2010). In particular, birds have been widely used as ecological models for the study of the alterations generated by anthropogenic noise in animals (Patricelli and Blickley 2006). Several types of responses to anthropogenic noise have been reported across literature, many of which have focused on the role of noisy urban sites. For instance, House Finches (Haemorhous mexicanus) have been shown to be able to modulate the minimum frequency of their songs in response to noise (Bermúdez-Cuamatzin et al. 2011). Great Tits (Parus major) have also been shown to switch to higher minimum frequencies when exposed to low frequency noise pollution (Slabbekoorn and den Boer-Visser 2006; Halfwerk and Slabbekoorn 2009; Slabbekoorn et al. 2012).
Avian vocal adjustments to urban noise can vary spatiotemporally. Such is the case of Spotless Starlings (Sturnus unicolor) and House Sparrows (Passer domesticus) that have been found to start singing before dawn to avoid urban noise (Arroyo-Solís et al. 2013). Also, European Robins (Erithacus rubecula) seem to avoid acoustic interference with urban noise by singing at night in sites that are noisy during the day (Fuller et al. 2007). Regarding spatial responses, some bird species tend to avoid anthropogenic noise due to the masking of their acoustic signals (McLaughlin and Kunc 2013). Studies have even provided evidence that supports the idea that urban noise plays an important role in the assemblage level. For example, a study from a Neotropical city found trends of avian assemblages singing earlier in noisier urban sites (Marín-Gómez and MacGregor-Fors 2019). Even declines in species richness, one of the best understood emergent properties of avian communities, have been reported for urban noisy areas (Carbó-Ramírez and Zuria 2011; Fontana et al. 2011; González-Oreja 2017; Perillo et al. 2017; De Camargo-Barbosa et al. 2020).
Given that previous studies have shown that birds can differentially respond to urban noise, ranging from not vocalizing while masking noise occurs to leaving noisy conditions, here we assessed the citywide relationship between anthropogenic noise levels and bird species richness in a Neotropical city. For this, we considered two urban noise measurements: (i) noise levels recorded during bird surveys (referred to as point-count noise hereafter) and (ii) daily average noise levels (referred to as 24 h noise hereafter). Given the available knowledge on the responses of birds to urban noise, we hypothesized two mutually exclusive outcomes. If birds are affected by urban noise in such a way that at least some of them avoid noisy sites, we predicted both noise measurements to be negatively related to avian species richness. Conversely, considering that vocalizations represent the most informative source when conducting point-counts (Ralph et al. 1995; Bibby et al. 2000), if some birds change their singing activities and become less detectable during surveys and/or momentarily fly away from noisy conditions, we expected a negative relationship between point-count noise levels and species richness, but no relation with 24 h noise. Finally, we also assessed the relationship between green cover and avian species richness, as green cover has been regarded as one of its main positive drivers in urban areas, and could contextualize our noise results (Fischer et al. 2012; Stirnemann et al. 2015; Schütz and Schulze 2015; Marzluff 2016).
This study was conducted in the city of Xalapa, state capital of Veracruz (Mexico; 19° 32ʹ 38ʺ N, 96° 54ʹ 36ʺ W; 1120–1720 m asl; INEGI 2009). Original vegetation in the region where Xalapa is settled was mainly comprised by montane cloud, tropical dry, and temperate forests (Castillo-Campos 1991). We located a total of 114 sampling sites following a citywide framework. For this, we considered the 106 study sites used in Escobar-Ibáñez and MacGregor-Fors (2016) and added eight additional sites along greenspaces to rise their representativeness in our sample (Fig. 1).
We conducted 10 min point-counts (50 m radius) at all 114 sampling sites from 6:00 to 10:00 h during the breeding season of 2019 (April 22 to May 6). We surveyed each location once and recorded all birds seen or heard [except overflying individuals; following Ralph et al. (1995)]. We decided to perform limited radius surveys to assure that all recorded birds were actively using the surveyed area. We located our survey sites at a minimum distance of 250 m from each other to assure that sampling sites did not spatially overlap with maximal recording distances reported in field manuals (Ralph et al. 1995; Bibby et al. 2000).
While point-counts were conducted, we measured point-count noise levels using a sound meter (B & K Precision model 732A; A-weighted scale, fast time 30–130 dB; frequency range: 31.5–8 kHz; resolution 0.1 dB). We recorded 72 noise level measurements during 3 min at each site with the sound meter mounted on a tripod at 1.5 m vertically-positioned. Afterward, we calculated the average, minimum, and maximum values of point-count noise for each survey site.
For 24 h noise levels we placed ARUs (Autonomous Recording Units, 16 SM4 and 6 SM3 song meters; Wildlife Acoustics Inc. ©, Maynard, MA) in 61 sites distributed across the city of Xalapa. Given that we could not place ARUs safely at all 114 points, we placed them at safe sites located across a gradient of urbanization density of Xalapa. We programmed ARUs to record for three consecutive days (03 June 2017 to 18 June 2017) using the following schedule during an entire day (24 h): a continuous long recording (75 min) during dawn and dusk periods, and 5 min every 15 min (i.e., 5 min on, 10 min off) during the remaining time periods. We used the same sound gain settings for ARU microphones, both left and right (~ 24 dB), to accomplish comparable soundscape recordings among sampled sites. Moreover, we automatically extracted noise levels from recordings in Kaleidoscope Pro following a batch procedure maintaining the same parameters (i.e., 60 s sample period and 0.0 dB adjustment). This procedure allowed us to retrieve noise measurements that are comparable across sample sites and thus reflect relative noise amplitude values in band frequencies from 19.7 to 2000.0 Hz; commonly used on noise studies (Merchant et al. 2015; Wildlife Acoustics 2019). The overall data set obtained consisted of 1425.6 h of recordings. Using this information, we calculated the mean amplitude for each site per day as a sample period of 1-min recording every 1 min, resulting in a total of 655 noise measurements. For purposes of this study, we used a global proxy of noise level, defined as the average noise amplitude per day at each study site considering the 1/3-octave band levels (Luther and Gentry 2013; Slabbekoorn 2013). We then calculated the logarithmic average of noise values because noise levels are on a logarithmic scale, as decibels increase exponentially. As noise levels were expressed as a relative measure (dB relative to 1 Volt), we could not convert them to sound pressure units (SPL). Thus, in our data set, noisier sites had values near 0 dB and quieter sites had values around − 100 dB. By using the logarithmic average values of noise, we generated a raster continuum of values that allowed us to retrieve 24 h noise levels in each one of the 114 sampling sites. We obtained rasterized values as result of an inverse distance weighting (IDW) interpolation, which estimates values using the nearest sample points available, which in turn, are weighted by a power proportional to the inverse of the distance between them and the desired value (Li and Heap 2008). Finally, we retrieved green cover values within 50 m radius buffers from all sampling sites using information of a satellite image classification [see Falfán et al. (2018) for further methodological details].
We performed a single linear model (LM) to assess potential relationships between the independent variables (i.e., point-count noise, 24 h noise, vegetation cover) with bird richness (dependent variable). Given that average, minimum, and maximum point-count noise values were correlated (r > 0.38, p < 0.001), we only considered maximum values, as they showed to have the highest statistical variance. We ran all statistical analyses in R (R Core Team 2019).
We recorded a total of 82 bird species, with average 13.19 (± SD 7.92) species richness per point-count. The most abundant species recorded was the Great-tailed Grackle (Quiscalus mexicanus, n = 409), followed by Social Flycatcher (Myiozetetes similis, n = 149), House Sparrow (n = 144), and Rock Pigeon (Columba livia, n = 130). As expected, our results show a positive relationship between vegetation cover and bird species richness (Table 1, Fig. 2).
| Variable | Estimate | SE | F | df | P |
| Point-count noise | − 0.0187 | 0.0276 | 17.269 | 1 | < 0.001 |
| 24 h noise | 0.0067 | 0.0261 | 0.011 | 1 | 0.918 |
| Vegetation cover | 0.0007 | 0.0001 | 76.889 | 1 | < 0.001 |
DownLoad:
CSV
The minimum point-count noise was 33 dBA, which is comparable to that of a library or a bedroom at night, while the maximum point-count noise was 97.6 dBA, similar to the sound made by a gas lawn mower or a diesel truck. The average point-count noise was 54.79 dBA, which corresponds to that of the noise produced in a commercial area (FAA 2018). In the case of 24 h noise levels, and considering that the scale is inverted (i.e., quieter sites are closer to − 100 dB, and noisier sites are closer to 0 dB; see Methods for further details), minimum 24 h noise level was of − 89.62 dB, while the maximum value was of − 44.52 dB, with an overall average of − 68.39 dB. Our records indicate that the noise recorded in our surveys is within the thresholds of a common city (Chepesiuk 2005; McAlexander et al. 2015; Kamenov 2016). We found a negative relationship between maximum point-count noise and avian species richness. Conversely, we did not find relationship between 24 h noise and bird species richness (Table 1, Fig. 2).
Birds that dwell within cities, including their greenspaces, are subject to numerous stimuli, pressures, and threats that vary spatiotemporally (Warren et al. 2006; Evans et al. 2011; Marzluff 2016; Santiago-Alarcon and Delgado-V 2017). In fact, the complex array of urban conditions and scenarios has been shown to mold the birds that are able to dwell within them (Melles et al. 2003; Evans et al. 2009; MacGregor-Fors and Schondube 2011; MacGregor-Fors and García-Arroyo 2017). As expected, our results showed a positive association between vegetation cover and bird species richness. This is consistent with the mounting evidence that urban vegetation plays a crucial positive role for urban-dwelling birds [see Marzluff (2016) for an updated review]. Empirical evidence has clearly shown that well-vegetated urban sites provide a wider array of conditions and resources that allow for many avian species that are not tolerant to the urban life to be present within cities (Croci et al. 2008; Evans et al. 2011; Sol et al. 2013). Actually, when a recent ecological study was contrasted with the historical list of birds of Xalapa, it was evident that most bird species of this biodiverse city are concentrated along its greenspace network and well-vegetated residential areas (González-García et al. 2014; Escobar-Ibáñez and MacGregor-Fors 2016).
Additionally, our findings are consistent with our second prediction (i.e., negative relationship between point-count noise levels and species richness, but no relation with 24 h noise). This finding is in partial agreement with previous studies that measured noise during or after surveys were performed and related it to bird species richness in urban areas (e.g., Carbó-Ramírez and Zuria 2011; Fontana et al. 2011; González-Oreja 2017; Perillo et al. 2017; De Camargo-Barbosa et al. 2020). Even studies performed in non-urban highways have found this relationship to hold, with noisier conditions associated with lower avian species richness (Herrera-Montes and Aide 2011).
Nevertheless, day-round noise (24 h noise) was not associated with variations in avian species richness in this study, suggesting that birds leave noisy sites or stop vocalizing in a short time-scale, like that of short ecological surveys (e.g., point-counts, transects). This result is particularly interesting, as previous studies have shown that birds can be importantly affected by prolonged, chronic noise (Habib et al. 2007; Leonard and Horn 2008; Blickley et al. 2012) and can also adjust their behavior in relation to noise in differing times of the day (e.g., Fuller et al. 2007; Gil et al. 2015; Lee et al. 2017). Thus, not finding a relationship between 24 h noise and bird species richness suggests that urban birds may be capable of tolerating anthropogenic noise more than shown in previous studies measuring focal noise during or close to survey times. Such response is in agreement with the growing amount of evidence that urban wildlife can be highly phenotypically plastic (Bonier et al. 2007). For example, several bird species have been shown to be quieter in noisy conditions, often avoiding vocalization overlap with anthropogenic noise (Fuller et al. 2007; Halfwerk and Slabbekoorn 2009).
Although our results clearly agree with our second prediction, there are some methodological limitations that need to be considered in future studies seeking to untangle bird richness patterns in different urban noise conditions. Most importantly, noise measurements reflecting their variability across the day should be taken in the exact same sites. Also, given that bird detection probability decreases in sites exceeding ~ 45 dBA (as noise reduces the distance and area where the acoustic signaling of birds can be perceived; Barber et al. 2010; Ortega and Francis 2012), field surveys should include the use of bird song recordings in order to assure that birds are not singing during noisy events, which can be easily identified in spectrograms using simple freeware (e.g., Audacity, Raven).
Results of this study support the hypothesis that decreases in urban bird species richness do not necessarily imply the permanent absence of species in the surveyed sites, shedding light on potential factors related to measuring noise while bird diversity surveys are performed. Thus, these findings suggest that birds could: (i) temporarily fly away from or avoid sites when noisy, (ii) stop vocalizing while noisy events are occurring to avoid their signals being masked, or (iii) be undetected due to our inability of recording them because of the noisy events. Thus, future studies could test if our findings are generalizable and which of the suggested scenarios are driving them.
ARU: Autonomous recording units; dB: Decibel; dBA: A-weighted decibels; FAA: Federal Aviation Administration; LM: Linear model; IDW: Inverse distance weighting; SPL: Sound pressure units.
We are deeply thankful to Rafael Rueda Hernández for his support in the field and to Eleanor Diamant for the English grammar revision of the final version. This research was supported by CONACYT through a financial aid as research assistant to COC-M (scholarship #771343), through a graduate scholarship to MG-A (scholarship #700755), and through project grants (CONACYT: 250910 and 251526) and a chair fellowship at CIIDIR (Cátedras CONACYT researcher number 1640; project number 1781) to JRSL. MG-A also acknowledges the support provided by the Doctorate Program of the Instituto de Ecología, A.C. (INECOL).
IM-F, MG-A, and OHM-G conceived the idea; OHM-G and COC-M lead the field work; COC-M, MG-A, and IM-F performed the analyses and lead the writing, with all authors contributing significantly to the final version. All authors read and approved the final manuscript.
The data sets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Not applicable.
Not applicable.
The authors declare that they have no competing interests.
Figures(8) / Tables(9)
| Variable | Estimate | SE | F | df | P |
| Point-count noise | − 0.0187 | 0.0276 | 17.269 | 1 | < 0.001 |
| 24 h noise | 0.0067 | 0.0261 | 0.011 | 1 | 0.918 |
| Vegetation cover | 0.0007 | 0.0001 | 76.889 | 1 | < 0.001 |
DownLoad:
CSV
Email Alerts
RSS Feeds