Our study was conducted in the Pampas grasslands of southeastern South America (SESA grasslands; Azpiroz et al., 2012) extending along the coast of Samborombon Bay (Fig. 1) and covering about 244,000 ha. This region is temperate with hot summers (December–March) and cool, wet winters (June–August), with a mean annual temperature of 15 °C. The annual rainfall is around 1000 mm; the mean monthly rainfall varies between 50 and 150 mm. Except for small forest patches of Celtis ehrenbergiana, all other vegetation is mainly herbaceous. Vegetation is arranged in a complex mosaic of herbaceous communities mainly determined by edaphic factors (subtle topography and soil salinity gradient), while latitudinal (regional) variation plays a secondary role (Vervoorst, 1967). Grasslands have evolved under light grazing conditions but were heavily disturbed after the arrival of Europeans in the 16th century with the introduction of domestic herbivores and, probably, frequent fires (Soriano et al., 1991). The original community was characterized by a ~60 cm grassland dominated by tussock grasses including various species of Nasella, Piptochaetium, Bromus and Paspalum (Vervoorst, 1967) and in the lowland areas that are affected by temporary water supplies was characterized by Paspalidium paludivagum and Leersia hexandra. The study area is an important wintering site for migrant neartic birds (Lanctot et al., 2002; Birdlife International, 2020b) and is listed as an Important Bird Area (IBA) of Argentina due to several neotropical migrant species as well as endangered grassland species such as the Dot-winged Rail Porzana spiloptera (Birdlife International, 2020c) and the Bay-capped Wren-Spinetail Spartonoica maluroides (Birdlife International, 2020d).

Fig. 1.

At the top, location of the study area within the area of the Flooding Pampas, Argentina; yellow asterisks represent the fields surveyed. Source: 2020 Google Earth, Cnes/Spot Images. At the bottom, images of the three treatments within the Flooding Pampas of Argentina where birds and vegetation were sampled: ‘‘Ryegrass promotion” (left), ‘‘Rotational grazing (middle); ‘‘Continuous grazing’’ (right).

(0.84MB).

We conducted surveys on ranches located in General Lavalle County, Buenos Aires province, where cattle-breeding and farming under traditional continuous grazing are the main activities. Livestock density is typically 0.7–1 head of cattle ha–1. We defined three management practices (treatments, see Fig. 1): (1) Grassland treated with the herbicide glyphosate (hereafter, Ryegrass promotion) that regularly received late-summer applications to promote winter forage production (i.e., Ryegrass). Glyphosate was applied with terrestrial spray equipment at a dose of four to five l/ha in a single application during the first week in March each year. At this time, warm-season species are growing actively and cool-season species (Ryegrass) start their growing period (see Rodriguez and Jacobo, 2010). This practice increased winter forage production per hectare and allowed improvement of stocking rate and meat production (Bilello and Zeberio, 2002); (2) Rotational grazing, grassland in which cattle were rotated among paddocks every 15–30 days during spring and summer and every 3–4 months during autumn and winter (see Jacobo et al., 2006), (3) Continuous grazing, grassland in which grazing was maintained throughout the year within the same paddocks (i.e., the modal system in the study area; see González-Fischer and Bilenca, 2020).

Avian assemblages and vegetation characteristics were evaluated in the three treatments. Our field work included three periods of the breeding cycle of the birds, early (the sampling was made in October), late (December) and post-breeding (late February/March), along two years (2011/2012 and 2012/2013). For each treatment we located systematically 12–15 count points, one per paddock (paddock size varied from 12 to 25 ha), separated from each other by at least 150 m to ensure their independence and separated by the same distance from fences, drainages and floodplains to avoid the edge effect. Surveys were conducted during the 3 h after dawn (0615 ± 10–0915 ± 10) and the 3 h before sunset (1710 ± 10v–v1950 ± 10). The same observer (MC) sampled each point count and recorded all birds seen or heard within a 50-m radius of the survey point during a 5-min period (following Bibby et al., 2000) excepting individuals in flight (Codesido et al., 2013; Agra et al., 2015). We used a handheld GPS (error ±5 m) to measure distances. We conducted one survey at each treatment by period (i.e., 35–45 point-counts per sample, thus a total of six samples and 247 point-counts over the study period, resulting in a sampling effort of 1235 min).

We determined avian species richness and abundance for each survey point for both SESA birds (Azpiroz et al., 2012) and non-SESA birds. SESA birds were grouped as short-grass species, broad species or tall-grass species following the criteria of Azpiroz et al. (2012). We followed Remsen et al. (2020) for taxonomy.

In addition, we also made a visual estimation of vegetation height and cover within each 50-m radius survey point (Matteucci and Colma, 1982) after completing the count of birds at each point. We divided each 50-m radius survey point into four quadrants according to the cardinal points (North, East, South and West), taking a measurement in each quadrant and finally averaging them to obtain an estimate per each 50-m radius survey point. According to the use of short grasslands by shorebirds (Isacch and Cardoni, 2011; Aldabe et al., 2018) and grassland birds (Agra et al., 2015) we classified vegetation height using a graduated ruler in the following three height strata: lower than 10 cm tall, 10–24 cm, and ≥25 cm. Then, the dominant plant species were identified, grouped according to functional groups (Lavorel et al., 1997) and the coverage of each group was identified and estimated for each of the height strata, considering the whole 50-m radius. In this way, each 50-m radius survey point had a different percentage of each cover stratum which, in turn, changes over time. We used Roitman and Preliasco (2012) to assign the grasses and forbs species to the following functional groups: Annual winter grasses, Legumes, Creeping summer grasses, Perennial winter grasses, Graminoids, Erect summer grasses, Forbs, Grass weeds, Dicotyledon weeds, and Erect summer dicotyledons.

Attributes of bird assemblages were analyzed by generalized linear mixed models (Zuur et al., 2009). The generalized linear mixed model provides a useful approach for analyzing repeated measures data by accounting for the lack of independence between observations that are repeated over time and to model more than one error term (Zuur et al., 2009). In this regard, each survey was considered a repeated measure. The response variables were bird species richness and abundance, and the explanatory variables were grazing management technique (3 levels: Ryegrass promotion, Rotational grazing and Continuous grazing) and period of the breeding season (3 levels: early, late and post-reproductive) as well as the interaction between both factors. Grazing management technique and period were fitted as fixed effects and survey point as a random term. Since the variance was much greater than the mean, both species richness and abundances of species data fitted negative binomial distribution and a log link function was used in all models (Zuur et al., 2009). Since no significant interannual variation was observed in the climatic variables, the year factor was not included in the model (the points between years were considered independent). Plots of residuals and normal probability were examined in order to evaluate assumptions of normality and homogeneity of variance by means of graphical validation tools (Zuur et al., 2009). The same analysis was used to compare vegetation cover (response variable) at different heights between grazing treatments. Vegetation cover data fitted beta distribution and a logit link function was used in all models (Zuur et al., 2009).

To identify bird species characteristic of each habitat under each management technique, we used an indicator-species analysis (De Cáceres and Legendre, 2009). This analysis calculates an indicator value for each species on the basis of its relative frequency and relative abundance in each category. Indicator values can range from 0 (no indication) to 100 (perfect indication). A perfect indication for a given habitat means that the species was recorded on all points within that habitat and was not observed in any other habitat. The function also allows evaluation of indicator species of both individual treatments and combination of treatments, as explained in De Cáceres (2013). We tested indicator values for significance with a Monte Carlo randomization procedure which compares the observed indicator values to alternative values calculated from the same data and randomly assigned to habitat type. We report only species whose indicator values are significant (P < 0.05).

Community composition were analyzed using Nonmetric Multidimensional Scaling (NMDS) with the Bray–Curtis index, we plotted the six samples for each grazing management type (3 different level of seasons by 2 years) in a multidimensional space defined by the total number of individuals of each bird species to graphically represent associations among samples.

Most of these analyses were implemented with R statistical software, Version 3.5.3 (R Development Core Team, 2019). For generalized linear mixed models we used glmmTMB package (Magnusson et al., 2017), and for indicator-species analysis we used indicspecies package (De Cáceres, 2013). For multidimensional scaling we used Past (Hammer et al., 2001).

The structure of the vegetation was different between grazing management treatments (Fig. 2). Vegetation cover – less than 10 cm – was greater in both Rotational and Continuous grazing sites in comparison to the Ryegrass promotion sites (Generalized linear mixed model, treatment effect: Tukey test, P < 0.05; Fig. 2a). For this cover stratum in both Rotational grazing and Continuous grazing we found a greater number of species and functional groups than in Ryegrass promotion sites, including Legume species (Lotus sp., Trifolium sp. Melilotus sp.), and other creeping summer grasses (Paspalum sp., Stenotaprhum secundantum, Distichlis spicata) of forage interest (Table 1), whereas Ryegrass promotion sites had Ryegrass as the dominant species and to a lesser extent some grass and dicotyledon weeds (Table 1). Vegetation cover at the 11–24 cm stratum was greater in Ryegrass promotion sites in comparison to the Rotational grazing sites (Generalized linear mixed model, treatment effect: Tukey test, P < 0.05; Fig. 2b). For this cover stratum, the Ryegrass was again the dominant species in herbicide application sites, while this species in addition to legumes and creeping summer grasses were the dominant species for Rotational grazing and Continuous grazing (Table 1). Finally, vegetation cover – greater than 25 cm – was similar among treatments (P > 0.05; Fig. 2c). Ryegrass promotion sites had the Ryegrass as the dominant species (Table 1), while Rotational grazing and Continuous grazing have this species in addition to perennial winter grass (Nasella neesiana, Jarava sp.) and legumes (Table 1).

Fig. 2.

Total vegetation cover at different height strata for grasslands with Ryegrass promotion, grasslands with Rotational grazing and grasslands with Continuous grazing during spring-summer 2011/2012 and 2012/2013 in the Flooding Pampas, Argentina. Different letters indicate significant differences (Tukey test; P < 0.05).

(0.29MB).

In total, we recorded 3853 individuals and 41 bird species, from which 27 were SESA species (6 short-grass species, 21 broad species and none tall-grass species) and 14 non-SESA species (Table 2).

We observed 29 species and 947 individuals in the Ryegrass promotion, 33 species and 1595 individuals under Rotational grazing, and 32 species and 1311 individuals under Continuous grazing (Table 2). In addition, 20 species that were detected in the Ryegrass promotion sites, 23 species detected in Rotational grazing sites and 23 species detected in the Continuous grazing sites are considered representative of southeastern South American grasslands. Among SESA species, the American Golden Plover and Buff-breasted Sandpiper, both nearctic migratory shorebirds, represented 31% of the total number of individuals seen. Other abundant SESA species included the Southern Lapwing (Vanellus chilensis) and Correndera Pipit (Anthus correndera), followed by the Brown-and-yellow Marshbird (Pseudoleistes virescens), Grassland Yellow-finch (Sicalis luteola), and the White-faced Ibis (Plegadis chihi) a non-SESA species (Table 2).

The NMDS showed a clear separation in structure of bird communities among the three grazing treatments (Fig. 3; NMDS stress = 0.19). The first ordination axis groups the Rotational and Continuous grazing sites and separated them from the Ryegrass promotion sites, while the second ordination axis separated Rotational grazing sites from Continuous grazing sites.

Fig. 3.

Nonmetric multidimensional scaling based on species’ abundance from 18 sample sites within three categories of grazing management (Ryegrass promotion, grasslands with Rotational grazing and grasslands with Continuous grazing) in the Flooding Pampas, Argentina. Symbols represent the mean and standard error of samples sites in each grazing treatment to enhance visual interpretation of the ordination (NMDS stress = 0.19).

(0.11MB).

The total abundance of birds was lower in Ryegrass promotion sites in comparison with Rotational grazing sites (Generalized linear mixed model, treatment effect: Tukey test, P < 0.05; Fig. 4). In particular, the abundance of short-grass species was also lower in Ryegrass promotion sites when compared to Rotational and Continuous grazing sites (Generalized linear mixed model, treatment effect: Tukey test, P < 0.001 and Tukey test P < 0.01). The abundance of broad grassland species was similar among treatments (P > 0.05). Species richness was similar among treatments for all periods of the breeding cycle of the birds for all species (P > 0.05; Fig. 4), and for short and broad grassland birds (P > 0.05).

Fig. 4.

Total abundance and total richness of birds in grasslands with Ryegrass promotion, in grasslands with Rotational grazing and in grasslands with Continuous grazing during spring-summer 2011/2012 and 2012/2013 in the Flooding Pampas, Argentina.

(0.23MB).

Indicator-species analysis identified four species with significant indicator values; three were associated with Rotational and Continuous grazing and one with Ryegrass promotion (Table 3). Three of the six species associated with short grasses are indicators for Rotational grazing and Continuous grazing, the American Golden Plover, Buff-breasted Sandpiper and Austral Negrito (Lessonia rufa), whereas indicator species for the Ryegrass promotion is the Eared Dove Zenaida auriculata, a non-SESA species (Table 3).

Domestic European cattle have modified the floristic composition and structure of the grasslands of the Pampas (León et al., 1984). This process has drastically accelerated since the end of the 19th century, when cattle became confined and therefore Continuous grazing initiated (Deregibus et al., 1995). Traditional continuous grazing has deteriorated these grasslands, leading to 1) severe reduction of winter grasses which are highly preferred and severely defoliated during winter, 2) invasion by exotic and native forbs of low forage value, 3) soil denudation, and 4) reduction of stocking rate, animal performance, and net profitability (Deregibus et al., 1995).

Environmental and grazing management problems related to traditional continuous grazing have promoted the search for alternative management practices, including those based on the use of agrochemical inputs and simplification of the natural system versus those that are based on knowledge of natural processes (Bilenca et al., 2018). Our results are in agreement with management studies showing that agrochemical inputs may be more detrimental for wildlife conservation than those which are based on understanding ecological processes (Eisler et al., 2014; Bilenca et al., 2018, and references therein). Application of glyphosate in the native grasslands of the Flooding Pampa has been promoted to increase winter forage production. However, this practice changes species composition, floristic richness, diversity of seed bank and greatly reduces the summer and annual aboveground net primary production (Rodriguez and Jacobo, 2013; Rodriguez et al., 2018). The recurrent spraying of glyphosate in the Flooding Pampa grassland negatively affects ecosystem functioning and risks biodiversity loss in this natural habitat (Rodriguez et al., 2018). In particular, grassland management with the herbicide glyphosate has a negative effect on the functional groups of vegetation and in the structure of grassland bird assemblages in the Flooding Pampas.

Rotational grazing has usually resulted in an immediate increase of the stocking rate of about 30% (Jacobo et al., 2006). The increase of species of high forage value, decline in lower foraging value species, and the reduction of bare soil improve rangeland conditions and carrying capacity (Jacobo et al., 2006) due to the increase of winter forage productivity (Jacobo et al., 2000). Previous work in the study area has shown the potential of Rotational grazing to successfully attend both productive goals as well as conservation of endangered herbivores like Pampas deer Ozotocerus bezoarticus (Rodriguez et al., 2016), as well as conservation of the biodiversity of native birds (Vaccaro et al., 2020). In our study, shorebirds were closely associated with both Rotational and Continuous grazing management, whereas Ryegrass promotion using herbicides was detrimental for shorebirds.

Both livestock production and conservation compete for land in the Pampas ecoregion. Most wildlife biodiversity studies have focused on conservation rather than balancing biodiversity and domestic animal production, constraining the possibilities of benefiting both wild and domestic animals (Neilly et al., 2016; Martinez Ortiz et al., 2017). In this sense, conservation practices and livestock production in this region should be considered together, and discussed among farmers, conservationists, and government agencies (Dotta et al., 2015; González-Fischer and Bilenca, 2020).