Grassland birds are declining in South America (Vickery et al. Reference Vickery, Tubaro, Cardoso da Silva, Peterjohn, Herkert and Cavalcanti1999), North America (Askins et al. Reference Askins, Chávez-Ramírez, Dale, Haas, Herkert, Knopf and Vickery2007), Europe (Donald et al. Reference Donald, Green and Heath2001) and other parts of the world (Butchart et al. Reference Butchart, Stattersfield, Bennun, Shutes, Akçakaya, Baillie, Stuart, Hilton-Taylor and Mace2004). The occurrence of threatened grassland birds is highly influenced by agricultural land-use (Murphy Reference Murphy2003, Donald et al. Reference Donald, Sanderson, Burfield and van Bommel2006) and main current causes of their decline are intensification of agricultural practices, habitat fragmentation, use of pesticides and increase in exotic and woody vegetation (BirdLife International 2008). Intensification of traditional rangeland management (i.e. annual burning and intensive grazing) has contributed in some cases to this decline, as it promotes habitat uniformity instead of heterogeneity (Fuhlendorf et al. Reference Fuhlendorf, Harrell, Engle, Hamilton, Davis and Leslie2006).
During the last 20 years, the remaining natural grasslands of north-east of Argentina have been increasingly managed for livestock grazing as a result of the spread of cultivated areas for soybean (Oesterheld Reference Oesterheld2008). Typical rangeland management techniques include cross-fencing, water management, use of herbicides and annual burning (Gibson Reference Gibson2009). Although fire was a natural disturbance in the north-east of Argentina, its frequency was lower (one fire every four years; Kunst and Bravo Reference Kunst, Bravo, Kunst, Bravo and Panigatti2003). At present, annual burning is widespread in this region because it has a low cost, eliminates woody vegetation that tends to invade the grassland when fires are suppressed, and allows the rapid growth of fresh grasses (Morello and Adamoli Reference Morello and Adamoli1974, Kunst and Bravo Reference Kunst, Bravo, Kunst, Bravo and Panigatti2003). A study of fire patterns in South America shows that they occur in all the remaining natural grasslands of north-east Argentina and that most of them occur during late winter and spring (Di Bella et al. Reference Di Bella, Jobbágy, Paruelo and Pinnock2006). These changes will have caused the decline in many species of birds that inhabit grasslands and the north-east of Argentina holds a high number of globally threatened grassland species, such as Strange-tailed Tyrant Alectrurus risora, Sharp-tailed Tyrant Culicivora caudacuta, Saffron-cowled Blackbird Xanthopsar flavus, Black-and-white Monjita Heteroxolmis dominicana and several species of Sporophila seedeaters (BirdLife International 2009).
Direct effects of fire on birds include destruction of nests, while indirect effects involve changes in vegetation which may favour some bird species over others. Although there are a large number of obligate grassland birds in the Neotropics (Vickery et al. Reference Vickery, Tubaro, Cardoso da Silva, Peterjohn, Herkert and Cavalcanti1999) and several of them are threatened (BirdLife International 2009), very few studies have analysed how they respond to fire. These studies have mainly focused on how fires modify the composition of bird communities, e.g. cerrado (Cavalcanti and Alves Reference Cavalcanti and Alves1997, Parker and Willis Reference Parker and Willis1997), Amazonian savanna (Cintra and Sanaiotti Reference Cintra and Sanaiotti2005), and southern pampas (Comparatore et al. Reference Comparatore, Martínez, Vassallo, Barg and Isacch1996, Isacch et al. Reference Isacch, Holz, Ricci and Martínez2004). One recent exception is an experimental study by Petry and Krüger (Reference Petry and Krüger2010) in highlands of southern Brazil which found that the Saffron-cowled Blackbird avoided habitats with tall grasses and bushes and forage more frequently in burned than unburned areas, suggesting that fire could be beneficial for this species. However, none of these studies have analysed in detail how burning affects the reproductive success of grassland birds and how individual birds respond to fire.
The Strange-tailed Tyrant Alectrurus risora is a tyrannine flycatcher that inhabits savannas and wet grasslands in the south of Paraguay and north-east of Argentina (Ridgely and Tudor Reference Ridgely and Tudor1994, Di Giacomo and Di Giacomo Reference Di Giacomo and Di Giacomo2004; Figure 1a). It is an obligate grassland bird that depends exclusively on this habitat for nesting and feeding (Vickery et al. Reference Vickery, Tubaro, Cardoso da Silva, Peterjohn, Herkert and Cavalcanti1999, Fitzpatrick Reference Fitzpatrick, del Hoyo, Elliot and Christie2004). During the last century, it suffered a 90% contraction in its original range in association with conversion of natural grasslands to agriculture (Di Giacomo and Di Giacomo Reference Di Giacomo and Di Giacomo2004). At present, it has the IUCN Red List category ‘Vulnerable’ and further declines in range and number are expected because suitable habitat is subject to current extensive agricultural intensification and afforestation (Collar et al. Reference Collar, Gonzaga, Krabbe, Madroño Nieto, Naranjo, Parker and Wege1992, BirdLife International 2009). The current extent of occurrence of the Strange-tailed Tyrant in Argentina mainly includes private lands where fire is used as a management tool. Moreover, the breeding season of the species starts in early September (Di Giacomo and Di Giacomo Reference Di Giacomo and Di Giacomo2004, Di Giacomo Reference Di Giacomo2010), a time of year when fires are frequent in its present range. Therefore, the effects of grassland burning on reproductive success need to be understood to identify the most appropriate conservation measures for the species.
In this paper, we describe the response to fire of a breeding population of Strange-tailed Tyrant in the province of Formosa, north-east Argentina, that was monitored from September 1996 to January 2009. First, we explored the effects of fire on number of nesting attempts and their outcomes through a retrospective study of four accidental fires. In addition, we made a prospective study after one experimental prescribed burning of 50% of our study plot and recorded the distribution of banded males and females and the number of nesting attempts and their outcomes in the unburned and burned plots from two years before to two years after the experimental prescribed fire.
We conducted our study at ‘Reserva El Bagual’ (hereafter REB), in the province of Formosa, Argentina (26º18′23″S, 58º49′25″W, Figure 1a). REB is a 3,300 ha private protected area where livestock grazing and agriculture have been excluded since 1985 (Di Giacomo Reference Di Giacomo, Di Giacomo and Krapovickas2005a). The main habitat of REB is an open savanna of humid chaco (Cabrera and Willink Reference Cabrera and Willink1980). Pyrogenic grasses such as Brazilian satintail Imperata brasiliensis and Andropogon lateralis dominate the grassland in well-drained terrains, while Paspalum spp. dominate in wet lowlands. At present, grazing within the reserve is done by a small population of Marsh Deer Blastocerus dichotomus (IUCN status ‘Vulnerable’). REB is an Important Bird Area (IBA) that supports a diverse avifauna of 336 species, including populations of 13 species of conservation concern, such as breeding populations of Strange-tailed Tyrant and Sharp-tailed Tyrant (Di Giacomo Reference Di Giacomo2005b).
Strange-tailed Tyrants have a polygynous mating system, with males holding close territories of 2–3 ha with two females on average (Di Giacomo Reference Di Giacomo2010). Females nest from mid-September to mid-January and have on average two nesting attempts per breeding season (range 1–4; Di Giacomo Reference Di Giacomo2010). Nests are built near the ground at the base or among the leaves of tall grasses (see Results). Nesting success is 0.23 on average, modal clutch size is three eggs and successful nests fledge on average two chicks. Parental care is provided exclusively by the female, who builds the nest, incubates the eggs and broods and feeds the chicks (Di Giacomo Reference Di Giacomo2010).
Data collection and analysis
We collected data since 1996 in a grassland area of 50 ha located on the border of a small stream (Figure 1b) where 5–10 males and 10–20 females have nested regularly since 1995. This grassland is dominated by the tall grass (0.8–1.2 m in height) Imperata brasiliensis (Andropogoneae). Between 1996 and 2008 we recorded 292 nesting attempts (i.e. nests with laying) and for 286 of them we recorded the apparent nesting success (i.e. proportion of successful nests). In 283 of these nests we recorded the species of the plant in which the nest was built, in 261 the height of the plant, and in 265 the height of the nest (distance from ground to top of nest).
In 2004, in a 30 ha plot within our study area, we put poles (2 m in height) every 100 m forming a grid of 30 squares of 100 × 100 m (300 × 1,000 m, Figure 1b). We used this grid to delimit the territories of banded males and females using the territory mapping technique (Bibby et al. Reference Bibby, Burgess and Hill1992). During 2004–2005 we captured males during the non-breeding season and females during the non-breeding and breeding seasons using mist-nets. We ringed 8 males and 17 females on the tarsus with a numbered aluminium ring and a unique combination of coloured plastic rings. From 2004 to 2008 we mapped all observed individuals in the grid every 7–14 days during the entire breeding season (from early September to late January). We considered that a male had a territory when it was recorded during four weeks or more (Vickery et al. Reference Vickery, Hunter and Wells1992). Within the grid we found nests by observing female nesting behaviour (Martin and Geupel Reference Martin and Geupel1993) or by flushing females during systematic nest searching. We visited nests every 2–3 days until the chicks fledged or the nest failed.
From two years before to two years after the experimental prescribed fire (2004–2008) we estimated the main reproductive parameters (clutch size, hatching success, chick survival and nest survival) of nests in B and U plots (n = 146). We considered that the clutch was completed when the number of eggs remained constant for at least two consecutive days. We determined hatching success as the number of hatchlings divided by the number of eggs present in the nest at the time of hatching, and chick survival as the number of fledglings divided by the number of hatchlings. We estimated nesting success using the daily survival rate (DSR) estimator provided by program MARK (White and Burnham Reference White and Burnham1999). From the original data set we excluded 11 nests that had insufficient data for nest survival analysis. We considered that a nest was successful if it fledged chicks, depredated if all the eggs or chicks disappeared between two consecutive visits and deserted if the eggs were cold to the touch for two consecutive visits and the female did not attend the nest.
Fire history at REB
From 1996 to 2009 there were five fire events at REB. Three fires were initiated on neighbouring ranches dedicated to livestock grazing (October 8–13 1997; September 23–25 1999; and April 1–5 2003) and one within the reserve by illegal hunters (March 14–16 2001). Fires in 1997, 2001 and 2003 were of medium intensity and afterwards there remained small patches of unburned grasses. However, in 1999 there was a fire of high intensity that totally eliminated the vegetation cover. In 2004, the managers of REB started a programme of fire management to protect the habitat of the breeding populations of Strange-tailed Tyrants and Sharp-tailed Tyrants. The programme included the construction of a firebreak system and the use of prescribed fires to control the biomass of grasses and to prevent the invasion of woody vegetation (Morello and Adamoli Reference Morello and Adamoli1974). One of the firebreaks divided our study plot into two plots of 15 ha (Figure 1b). In August 2006 one of these plots was burned under controlled conditions (Marino Reference Marino2007). We used this experimentally burned plot and the unburned plot to study how fire affected the distribution of Strange-tailed Tyrants and their reproductive parameters.
For most analyses we used nonparametric statistics due to lack of normality of the data and relatively small sample size (Siegel and Castellan Reference Siegel and Castellan1988). We used Mann-Whitney U tests to compare the height of plants and nests between the years with and without fires. We used Kruskal-Wallis tests to compare reproductive parameters in the unburned plot between the years before the prescribed fire, the year of the prescribed fire and the years after the prescribed fire. We used a binomial test to examine if the distribution of males, females and nests between unburned and burned plots occurred with equal probabilities. We estimated specific daily survival rates with software MARK (White and Burnham Reference White and Burnham1999) and we used the program CONTRAST to examine differences in daily survival rates (Hines and Sauer Reference Hines and Sauer1989). All values reported are means ± SE.
Nesting attempts in years with and without fires
Figure 2 shows the number of nesting attempts per year in the 30 ha study plot and the time when fires occurred. Fires differed in origin, time of the year at which they occurred, size and intensity. These differences were probably the cause of the high variation in number of nesting attempts in years with a fire (range 2–30). The most noticeable effect of fire on the number of nesting attempts occurred in 1999. This year the fire occurred at the beginning of the spring (23–25 September) after an extremely dry winter. The fire eliminated the vegetation cover in almost all the study area. That year there were two nesting attempts only, and both occurred at the end of the breeding season (12 December and 5 January) in two small patches that had not been affected by the fire. In the other three accidental fires and in the prescribed fire, we did not observe a clear decrease in the number of nesting attempts, probably because the birds nested in the remaining patches of unburned grassland. As a result of this, we did not detect differences in the number of nesting attempts between years with and without a fire (with: 20.2 ± 4.9, n = 5; without: 24.1 ± 3.2, n = 8; Mann-Whitney U test: Z = −0.59, P = 0.56).
Plant and nest heights in years with and without a fire
Sixty-two percent of the nests (175/283) were built in Imperata brasiliensis, the dominant grass species in our study area. Other species of grasses used for nesting were Paspalum spp. (23/283), Andropogon spp. (13/283), Eustachys distichophylla (12/283), Elionurus muticus (11/283) and Schizachyrium spicatum (10/283). The frequency of use of Imperata brasiliensis was lower in years with a fire than in those without a fire (22/68 vs. 153/215; χ 2 = 31.3, P < 0.0001).
The height of the plant used for nesting was lower in years with a fire than in those without a fire (54.9 ± 2.6 cm, n = 60 vs. 80.7 ± 1.4 cm, n = 201; Mann-Whitney U test, Z = 7.4, P < 0.0001; Figure 3a). Similarly, in years with a fire the nests were built at a lower height than in years without a fire (6.0 ± 2.6 cm, n = 59 vs. 40.0 ± 1.4 cm, n = 203; Mann-Whitney U test, Z = 8.9, P < 0.0001), with nearly 80% of the nests built directly on the ground (Figure 3b). We did not detect differences in apparent nest survival between years with a fire and without a fire (with fire: 18/68; without fire 62/156; Contingency test χ 2 = 0.03, P = 0.87).
Distribution of males, females and nests before and after the prescribed fire
During the breeding seasons 2004 and 2005 (before the prescribed fire) there were seven and eight males and 17 and 16 females, respectively. The number of nesting attempts during these breeding seasons was 38 (2004) and 23 (2005) (Figure 4). The settlement of males and females and the number of nesting attempts in plots B and U did not differ from random (binomial tests, P > 0.13 for all comparisons, Figure 4). During the 2006 breeding season (year of the prescribed fire) all males and females settled in the unburned plot and all nesting attempts occurred in that plot (Binomial tests, P = 0.03 for males, P < 0.001 for females and P < 0.001 for nesting attempts, Figure 4). In 2006, the number of males in the unburned plot was double the number observed in the same plot during the previous breeding seasons. Moreover, it held three ringed females that had nested in the burned plot the previous breeding season. The other five ringed females that nested in the unburned plot had nested there the previous season. During the breeding season 2007 (one year after the prescribed fire) there were two males, three females and six nesting attempts in burned plot, while in the unburned plot there were three males, eight females and 17 nesting attempts (Binomial tests, P = 0.50 for males, P = 0.11 for females and P = 0.02 for nesting attempts). During this breeding season, one banded female that had switched from the burned to the unburned plot during the breeding season 2006 returned to the burned plot. Finally, during the breeding season 2008 the number of females and nesting attempts in the burned plot increased from 3 to 8 and from 6 to 14, respectively (Binomial tests, P > 0.30 for all comparisons, Figure 4).
Reproductive parameters and nesting success before and after the prescribed fire
Daily survival rate in the unburned plot the year of the prescribed fire did not differ from that of years before the fire (χ 2 = 0.15, P = 0.70, Table 1). However, DSR in the unburned plot after the prescribed fire was higher than that of the year with fire (χ 2 = 5.76, P = 0.016), or the years before fire (χ 2 = 6.73, P = 0.01). Daily survival rates in the burned plot did not differ between years before and after the prescribed fire (2004–2005: 0.960 ± 0.010; 2007–2008: 0.964 ± 0.009; χ 2 = 0.09, P = 0.77, table 1).
Clutch size, hatching success and chick survival in nests built in the unburned plot did not differ between the years before the prescribed fire, the year with the fire and those after the fire (Kruskall-Wallis test, H 2 = 2.87, P = 0.24 for clutch size; H 2 = 3.18, P = 0.20 for hatching success; and H 2 = 0.63, P = 0.73 for chick survival). As a result, there were no differences in the number of young fledged per nest between the years before the fire (2004–2005: 2.29 ± 0.33, n = 7), the year with the fire (2006: 2.00 ± 0.35, n = 6) and those after the fire (2007–2008: 2.21 ± 0.23, n = 14, Table 1).
Our results show that grassland burning affected the reproductive behaviour of Strange-tailed Tyrants. In years with a fire, the height of the plant used for nesting and the height of the nest was lower than in those without a fire, but this was not associated with changes in apparent nest survival. Data on female distribution before and after the prescribed fire show that all females avoided nesting in the burned plot in the breeding season immediately after the fire. Females started to return to this plot in the second breeding season after the fire and by the third breeding season they did not discriminate between burned and unburned plots. Consistent with this observation, Albute (Reference Albute2009) reported that one year after the fire, above-ground biomass was significantly lower in the burned than in the unburned plot, but after two years it did not differ between plots. The higher density of nests in the unburned plot in the year of the prescribed fire did not affect the main reproductive parameters of Strange-tailed Tyrants, as there were no differences in nest survival, clutch size, hatching success and chick survival between that year and the years before and after the fire. In addition, nest survival in the burned plot in the years after the fire did not differ from that in the years before the fire.
Fire could be a useful tool for managing the habitat of obligate Neotropical grassland birds such as Strange-tailed Tyrants because it allows the maintenance of grasses by preventing the increase of woody cover (Morello and Adamoli Reference Morello and Adamoli1974). Bush encroachment has been proposed as an important threat for the Sidamo Lark Heteromirafra sidamoensis, a ‘Critically Endangered’ grassland endemic in southern Ethiopia, and reversing fire suppression policies appears necessary to avoid the extinction of this species (Spottiswoode et al. Reference Spottiswoode, Wondafrash, Gabremichael, Dellelegn Abebe, Mwangi, Collar and Dolman2009). Likewise, based on a population viability analysis Duca et al. (Reference Duca, Yokomizo, Marini and Possingham2009) recommended the use of fire management in small reserves in cerrado savannas for conserving a ‘Near Threatened’ endemic, the White-banded Tanager Neothraupis fasciata.
Prescribed fire could be a management tool for maintaining suitable habitat for Strange-tailed Tyrants in pyrogenic grasslands (i.e. to avoid bush encroachment) and for reducing the risk of large accidental fires (like that of 1999), which can suppress breeding and threaten the population. Our results indicate that the impact of prescribed fires on the reproduction of Strange-tailed Tyrant could be minimised by choosing the appropriate season at which the grassland is burnt (autumn or early winter), the size and intensity of the fire, and its frequency. In years when accidental fires did not affect the whole area (1997, 2001 and 2003) the birds nested in the remaining unburned patches and the number of nesting attempts was similar to years without fire; and although in these years Strange-tailed Tyrants nested in lower plants and at a lower height, this was not associated with a lower nesting success.
As regards the frequency of fires, we observed that the third breeding season after the prescribed fire the birds did not show avoidance of the burned patch, suggesting that it had by then become as attractive as the unburned one. Similarly, Petry and Krüger (Reference Petry and Krüger2010) commented that Saffron-cowled Blackbirds returned to a burned marshland to breed only when the plants were developed enough and the marshland structure was recovered, which occurred three breeding seasons after the fire. These results suggest that the intervals between burns should be longer than two years. However, our study cannot provide an indication of how long the interval should be, as we did not have long periods without fire. The burning of small fragments of grasslands would allow a mosaic of unburned nearby patches to be maintained as refuges, thus facilitating the colonisation of burned patches once the vegetation recovers. In addition, some authors have reported that the availability of herbivorous arthropods increases in recently burned patches (Warren et al. Reference Warren, Scifres and Teel1987, Swengel Reference Swengel2001, Shochat et al. Reference Shochat, Wolfe, Patten, Reinking and Sherrod2005), which could provide Strange-tailed Tyrants additional food, although this requires further investigation.
On the contrary, burning of the grassland every year, the common practice for livestock grazing in the region (Kunst and Bravo Reference Kunst, Bravo, Kunst, Bravo and Panigatti2003, Di Bella et al. Reference Di Bella, Jobbágy, Paruelo and Pinnock2006), has a negative impact on Strange-tailed Tyrant settlement for reproduction. The same effect was observed in other grassland birds where the frequency of fires notably affects the abundance of grass-dependent species, which are least abundant or absent at sites in the breeding season immediately following burning (Powell Reference Powell2006). For example in USA, Henslow’s Sparrow Ammodramus henslowii, a grassland obligate of conservation concern, is completely absent from pastures that are annually burned (Coppedge et al. Reference Coppedge, Fuhlendorf, Harrell and Engle2008). Similarly, in Australia, detrimental fire regimes contributed to the extinction of two of the three bird species and three of the four bird subspecies which have disappeared since European colonisation (Woinarski and Recher Reference Woinarski and Recher1997).
In north-east Argentina there are 16 Important Bird Areas (IBAs) that include populations of Strange-tailed Tyrants (Di Giacomo Reference Di Giacomo2005b). These IBAs comprise approximately 1 million ha of wet grasslands and savannas. Only six of these 16 IBAs (covering 150,000 ha) are managed to some extent as protected areas with the exclusion of livestock and under some type of fire management. The remaining IBAs are private land dedicated to livestock grazing where the commonest management practice is annual burning. Under this scenario, it would be expected that Strange-tailed Tyrants and other threatened species that require tall grass habitat for breeding have experienced reductions in their populations or local extinctions. Such species of birds are now at increasing risk of extinction because of intensifying livestock management in response to spreading soybean cultivation (Oesterheld Reference Oesterheld2008). If we want to conserve viable populations of Strange-tailed Tyrants, we urgently need to influence the current use of fire on private lands dedicated to livestock grazing where this species is still present.
We thank Alparamis SA and Aves Argentinas for allowing us to conduct this study at ‘Reserva El Bagual’ and Dr. Gustavo Marino for helpful comments on the ecology of subtropical grasslands. ASDG was supported by a Canon National Park Science Scholarship and a Doctoral Scholarship of the University of Buenos Aires. JCR is Research Fellow of the Consejo Nacional de Investigaciones Científicas y Técnicas de Argentina.