Introduction

Strongly seasonal environments and regions face dramatic annual and interannual variations in their conditions, resources, and communities (Gaston 2000; Bunn et al. 2006; Tonkin et al. 2017; Whitford and Duval 2019). This might lead to the occurrence of biogeographical regions and sub-regions of arid or semi-arid environments (e.g. Bone et al. 2015; Dexter et al. 2015). Among them are the Seasonally Tropical Dry Forests occurring in the Neotropics (Sarmiento 1975; Pennington et al. 2009; Silva and Souza 2018). These biomes harbor a high diversity of species and ecosystems, and considerable portions of their extensions still require multi-taxa inventories for appropriate knowledge and conservation of their biodiversity (Banda et al. 2016; Silva et al. 2017). This includes the avian fauna found in the Caatinga (Araújo and Silva 2017).

Caatinga is the dry forest region that lies in northeastern Brazil (Andrade-Lima 1981; Leal et al. 2003; Silva et al. 2017). This eco-region harbors a wide range of wildlife, including a diverse avifauna with near 550 species (Leal et al. 2003; Araújo and Silva 2017; Silva et al. 2017). Caatinga has been considered an important center of endemism in South America, harboring about 20 endemic bird species (Cracraft 1985; Haffer 1985; Stotz et al. 1996).

Birds have been studied in the Caatinga since the 1800s, when naturalists visited this Brazilian region to collect specimens for overseas museums (Pacheco 2002). This review showed that overseas scientists have produced most of the knowledge on Caatinga’s native birds until the early 1900s. Since then, researchers of some Brazilian museums have been responsible for most bird collects. Through several expeditions, they dominated the ornithological investigations in the Caatinga until the 1980s, while avian inventories involving mainly captures and records by sight or vocalizations became more frequent since the 1990s (Pacheco 2002; Silva et al. 2003; Araújo and Silva 2017). Thus, inventories published since 1990 have involved a set of techniques and equipment, not prioritizing collects as occurred in previous studies.

Therefore, inventories published in the last three decades differ markedly from those published until the late 1980s, in relation to the range of methods used to sample bird communities. They have been done inside and outside conservations units through the Caatinga (Araújo and Silva 2017). For example, large reserves have been studied on its western portion (e.g. Olmos 1993; Olmos and Albano 2012; Silveira and Santos 2012). Additionally, the Caatinga eastern portion was investigated by numerous studies (e.g. Kirwan et al. 2001; Las-Casas et al. 2012; Sousa et al. 2012; Las-Casas et al. 2019). Other inventories have sampled its northern portion (e.g. Silva et al. 2012; Bezerra et al. 2013; Santos et al. 2013), while some investigations studied the southern and central regions (e.g. Parrini et al. 1999; Dornellas et al. 2012; Nunes and Machado 2012; Mariano 2015, 2016).

The number of inventories of Caatinga’s birds has increased mainly after major publications about its biodiversity (e.g. Pacheco 2002; Leal et al. 2003; Silva et al. 2003) and the publication of a special issue in the Brazilian Journal of Ornithology (Revista Brasileira de Ornitologia), in 2012. More recently, Mendes and Sousa (2016) elaborated a compilation of some aspects of the methods and results of 19 avian inventories carried out in the Caatinga, but they have not provided an analysis nor a discussion regarding this information. Thus, aspects of the methods employed in more than 50 inventories of Caatinga’s bird species published since the 1990s remain not examined in detail. Reviews of the aspects of methods employed in inventories can bring information useful for researchers, managers and policy-makers.

The objective of this study was to elaborate a review of aspects of the methods adopted in recently published inventories of bird species conducted in the Caatinga. First, I reviewed information related to: 1) the authorship and home of these publications, 2) the location of the field activities in the Caatinga region, 3) the decades, years, months, and seasons in which the samplings occurred. Additionally, I reviewed the techniques adopted to detect birds and the equipment used to obtain and document records. Aspects of the sampling effort and of the estimation of species richness also were evaluated. Further, I examined the destination of the vouchers (skins, sound recordings, and photographs) that supported record documentation. To conclude this review, I analyzed the whole information and provided suggestions for future avian inventories in the Caatinga.

Methods

The Caatinga

Caatinga is the largest South American dry forest region, covering approximately 800,000 km2 exclusively in Brazil (Andrade-Lima 1981; Leal et al. 2003; Silva et al. 2017). It is bordered by the marine environment along most of its northern portion, by the Atlantic Forest on its eastern portion and by the Cerrado along its western and southern regions (Ab'Saber 1977; Rizzini 1997). The Caatinga environment is marked by high temperatures, irregular rainfall, and landscapes covered by xeric vegetation that includes a range of forests and shorter vegetation (Prado 2003).

Criteria considered in this review

Inventories examined in this review were those conducted within the Caatinga’s limits (sensu Leal et al. 2003), including those areas covered by humid forests (brejos de altitude) and Cerrado vegetation. Thus, publications such as Roda and Carlos (2004) and Pichorim et al. (2014) were also evaluated. One publication (one avian inventory) was considered a sample in this review. Thus, if a given inventory sampled the communities of several localities (e.g. Albano and Girão 2008; Lyra-Neves et al. 2012; Pereira et al. 2014) the sampling effort considered for this study corresponded to the total effort applied by field workers in the whole set of localities.

Studies focusing the ecology and/or structure of bird communities also were reviewed when substantial attention was given to the inventory of the localities, including the presentation of checklists of the recorded species (e.g. Santos 2008; Pereira and Azevedo-Junior 2011; Las-Casas et al. 2012; Lima et al. 2019). Studies of communities focusing aspects other than their inventorying, and that have not shown data regarding the whole studied community (e.g. Araújo et al. 2017; Gonçalves et al. 2017a, b) were not included in this review, despite their substantial contribution to the knowledge of the Caatinga’s avifauna. These selections and cut-offs were done to review only studies that focused the inventory of communities – knowing what species occur in a given region, area or locality.

When elaborating the map of the geographic distribution of localities/regions sampled by the examined inventories (Fig. 1), some procedures were taken. First, when a study inventoried a given area, such as a conservation unit (e.g. Olmos 1993; Santos et al. 2013) or a group of non-protected localities (e.g. Olmos et al. 2005), I considered as geographic coordinates of these studies the latitude and longitude of the median locality within the range of sampled sites. This was done because not all localities studied by investigators had their exact position informed, and because an inventory was considered as a study in this review. When a few studies sampled localities within an extensive region, I considered as distinct those localities distant by more than 100 km. For example, the 13 localities sampled by Silva et al. (2012) in Rio Grande do Norte appear as three points in the map. Thus, the number of localities present in the map (Fig. 1) is smaller than the total number of localities actually studied by investigators. Also regarding the map, I classified inventories into three categories, based on the information regarding the numbers of days involved in their sampling efforts (Fig. 1). Studies that provided this information were divided in two groups: those with ≤ 40 days of fieldwork, and those with > 40 days of fieldwork). This division at the “40 days” was adopted because most studies involved 40 or less days of fieldwork (see Section “Number of days of field activities”).

Fig. 1
figure 1

Geographic distribution of localities/regions where avian inventories published in the last three decades (1999–2019) were conducted in the Caatinga. The colors of circles indicate the number of days involved in the sampling effort of each inventory/study: the number of days was not informed (white circles), the sampling lasted for 1 ̶ 40 days (orange circles), the sampling lasted for more than 40 days (black circles)

Full size image

Literature review

Inventories of bird species carried out in the Caatinga, and published between 1990 and 2019 were included in this review. I have not included previous inventories because they basically involved only collecting (Pacheco 2002; Silva et al. 2003), and not a range of methods involved in studied published in the last three decades. Their search was based on the following sources of information: 1) the databases of Web-of-Science and Scopus; 2) Google Scholar; 3) major reviews on Caatinga and Brazilian birds (Oniki and Willis 2002; Pacheco 2002; Silva et al. 2003; Araújo and Silva 2017); 4) the curriculum vitae of several investigators of the Caatinga’s bird fauna available at the CNPq’s Plataforma Lattes; 5) the “References” section of publications found previously through these four sources of information. Besides examining peer-reviewed articles published in scientific journals, this review also included book chapters, and dissertations associated with Post-Graduation Programs of Brazilian Universities and museums.

Results and discussion

Publications and authorship

This review examined a total of 51 inventories published between 1990 and 2019 through three major ways of communication. Most (n = 47; 92%) of them were published in Brazilian or overseas scientific journals. The official journal of the Brazilian Ornithological Society (Ararajuba, and later as Revista Brasileira de Ornitologia), Atualidades Ornitológicas and the CEMAVE’s journal (Ornithologia) housed most (76%) of the studies examined – 23, 11 and 5 publications, respectively.

Other eight studies encompassed 16% of the inventories, and were published in journals such as Cotinga, Bird Conservation International and Zoologia. Two inventories were published as book chapters and other two in dissertations associated with Brazilian Universities (see References). Importantly, about a quarter of the reviewed inventories were housed in a “Caatinga” special issue of Revista Brasileira de Ornitologia in 2012, highlighting the importance of this journal in joining and publishing ornithological research about this Brazilian region.

The number of observers (field workers) per study ranged substantially within the set of inventories (Suppl. Mat. Table S1). Six inventories involved only one field worker. Studies counting with the participation of one, two, three, or four field workers, occurred in equal numbers (10 each, n = 40). Further, nine investigations had the participation of five to seven observers. Studies with more contributors were Vasconcelos et al. (2012), and Schunck et al. (2012), with 10 and 14 observers, respectively. Thus, most (78%) inventories of Caatinga’s bird species published since 1990 involved the recording of bird species by one to four field workers, while a minor portion counted with the effort of more observers (Suppl. Mat. Table S1).

The participation of several observers in a given inventory might bring advantages such as: 1) increasing the probability of obtaining high quality data in the Caatinga’s harsh environment, 2) sharing the time to be employed in the fieldwork, 3) the acquisition of more safety during activities in the study area, 4) the obtention of a larger sample of the communities, and 5) the opportunity of counting with the experience and opinion of other ornithologists during all phases of the study.

Distribution of recently published inventories within the Caatinga

Avian inventories published in the last three decades occurred in all states that harbor Caatinga landscapes (Fig. 1). Studies involving larger samplings (> 40 days of fieldwork) are spottily distributed throughout this dry forest region, but have not been conducted in its central portion. On the other hand, inventories regarding 40 or less days of sampling occurred in more localities, and are concentrated in two major regions: 1) northern Caatinga, 2) central and northern Bahia, and localities close to its northern limits. Further, inventories that have not informed the number of days involved in their sampling efforts comprised numerous localities that are mainly concentrated within an extensive region ranging from northern Bahia and western Sergipe to northern Caatinga (Fig. 1). As a result, major regions not or poorly sampled by recent avian inventories are located in Caatinga’s portions of southern Ceará, middle Piauí, western and eastern Bahia, and northern Minas Gerais. These areas should be considered as major targets of future inventories.

Besides this, several regions within the Caatinga still require inventories marked by large sampling effort. The states of Minas Gerais, Ceará and Alagoas still have not been inventoried by recent studies involving more than 40 days of field activities (Fig. 1). It is important to highlight that my review has not included inventories published prior to the end of the 1980s. These former studies, compiled in Pacheco (2002) and Silva et al. (2003), and unpublished records deposited in museums, archives, and databases should be considered for the elaboration of a complete figure on the geographic distribution of avian inventories conducted in the Caatinga.

Sampling inside and outside conservation units

Caatinga’s landscapes located inside and outside conservation units were sampled by investigators (Suppl. Mat. Table S1). Fifteen (29%) studies were based on records obtained exclusively within the limits of conservation units (e.g. Olmos 1993; Freitas and Borges 2011; Silveira and Santos 2012; Sousa et al. 2012). They involved samplings at five national parks by seven studies, two ecological stations by three studies, and one biological station by one study. Among these fifteen inventories were those that sampled communities of one national forest, two natural monuments, two state parks, and four RPPNs - private reserves of the natural heritage (Suppl. Mat. Table S1).

Other 29 (57%) studies investigated the avifauna only outside conservation units (e.g. Olmos et al. 2005; Farias 2007; Sousa 2009; Araújo and Rodrigues 2011; Freitas et al. 2013). On the other hand, only seven (14%) studies elaborated checklists of bird species by sampling areas inside and outside conservation units (e.g. Nunes and Machado 2012; Olmos and Albano 2012; Bezerra et al. 2013). These studies inventoried avifaunas in non-protected areas and in two ecological stations, three national parks, and one natural monument (Suppl. Mat. Table S1).

The availability of information regarding bird communities found both inside and outside conservation units is essential for the understanding of the occurrence of species within the Caatinga. For example, the conduction of avian inventories within a conservation unit is an important contribution to the knowledge of its role in protecting the regional biodiversity. These inventories also make possible investigations of the representativeness of the Caatinga’s avifauna in the system of conservation units, and the examination of beta-diversity through this biome. Further, comparisons of bird communities found in protected and non-protected areas can bring important information relative to the influence of landscape modification on the diversity of birds in the Caatinga (e.g. Araújo and Silva 2017). Also, comparisons of the avifaunas found within a conservation unit and in its surroundings might bring information regarding the needs or benefits of the expansion of its limits or the creation of further reserves nearby. Further, sampling non-protected landscapes can identify priority areas for the creation of conservation units (MMA 2002). Therefore, samplings carried out both inside and outside conservation units are valuable for the knowledge and conservation of the Caatinga’s avifauna.

Decades in which the studies were published

Two inventories published during the 1990s studied bird species at Parque Nacional da Serra da Capivara and in the Chapada Diamantina region (Olmos 1993; Parrini et al. 1999). About 27% of the reviewed inventories (n = 14) were published during the 2000s (e.g. Nascimento et al. 2000; Farias et al. 2005; Telino-Junior et al. 2005). On the other hand, a higher number of studies (n = 35) were published during the 2010s, comprising about 69% of the inventories evaluated in this review (e.g. Dornellas et al. 2012; Schunck et al. 2012; Barnett et al. 2014; Las-Casas et al. 2019) (Suppl. Mat. Table S1).

Therefore, the number of inventories conducted in the Caatinga has increased substantially during the last three decades. This growing production of articles results, in part, of recommendations made in important meetings and major publications regarding the Caatinga’s biodiversity (e.g. MMA 2002; Silva et al. 2003; Porto et al. 2004). Another important contribution was the publication of several studies in the issue 20(3) of Revista Brasileira de Ornitologia in 2012.

Number of years of the samplings

Three studies made no mention on the years in which bird communities were sampled (e.g. Roda and Carlos 2004; Lyra-Neves et al. 2012), while other two inventories informed only the first and the last years of a multi-year sampling (Albano and Girão 2008; Silva et al. 2012). On the other hand, most (n = 46; 90%) inventories pointed out the years when field activities occurred (Suppl. Mat. Table S1). They involved samplings within one to nine years (Fig. 2). Most (59%) of them studied communities within one or two years (e.g. Farias 2007; Roos et al. 2006; Santos 2008; Ruiz-Esparza et al. 2012). On the other hand, near a third of these inventories distributed the sampling effort along three to five years (e.g. Parrini et al. 1999; Nascimento et al. 2000; Kirwan et al. 2001). Further, a small proportion of the studies sampled communities within more than six years (e.g. Olmos and Albano 2012; Santos et al. 2013; Pereira et al. 2014) (Fig. 2).

Fig. 2
figure 2

Number of years involved in the samplings of 46 avian inventories conducted in the Caatinga and published in the last three decades (1990–2019). The other five inventories examined by this review have not provided information on this aspect of the sampling effort

Full size image

It is important to highlight that this approach of counting the number of years does not imply in the conduction of samplings during the 12 months of a given year. For example, if authors sampled communities in March of a given year, and in November of the following year, it was considered that samplings occurred in two years, not meaning that field activities occurred along 24 months.

Number of months involved in the samplings

Seven (14%) studies have not provided information about the number of months involved in their samplings in the Caatinga (Suppl. Mat. Table S1). When considering the other 44 studies (Fig. 3), there are three inventories that made samplings during a single month (Farias 2007, 2009; Santos et al. 2012), and relatively high number of investigations (n = 16) that examined communities within two or three months (e.g. Olmos et al. 2005; Farias et al. 2010; Paixão 2012). Further, comparable numbers of studies sampled avifaunas in four to eight months, such as those by Parrini et al. (1999) and Lima et al. (2003). In other four studies, samplings lasted for twelve months (e.g. Ruiz-Esparza et al. 2012; Las-Casas et al. 2012; Pereira et al. 2014) (Suppl. Mat. Table S1, Fig. 3).

Fig. 3
figure 3

Number of months involved in the samplings of 44 inventories of bird communities conducted in the Caatinga and published in the last three decades (1990–2019). The other seven inventories examined in this review have not provided information on this aspect of the sampling effort

Full size image

As a result, it is noted that 37 studies (73%) inventoried bird communities within twelve or fewer months (Fig. 3). Among the other seven studies were those conducted by Araújo et al. (2012) and Olmos and Albano (2012), that sampled bird communities during the highest numbers of months (Suppl. Mat. Table S1). Thus, their studies were more appropriate to record birds in different seasons of the year, a fact not achieved by 20 inventories that sampled communities in 1–3 months.

Number of days of field activities

Nineteen (37%) of the reviewed inventories have not informed the number of sampling days involved in their field activities (Suppl. Mat. Table S1). This percentage reached 50% of the studies, considering only those published during the 1900s. This same value occurred when only publications along the 2000s were considered. This lack of information occurred in 31% of the inventories, when considering only those published in the 2010s (Suppl. Mat. Table S1). Thus, the number of sampling days has been more often reported in more recent publications.

Inventories that informed this aspect of the sampling effort (n = 32; 63%) involved four to 124 days of field activities (Suppl. Mat. Table S1, Fig. 4). The two most frequent samplings were those of 1–20 days and 21–40 days. Together, they represented 66% of these 32 inventories. Additionally, intermediate sampling efforts (41–80 days) were performed by a set of eight studies (e.g. Las-Casas et al. 2012; Nunes and Machado 2012; Ruiz-Esparza et al. 2012), while only three studies sampled communities for more than 80 days (Araújo et al. 2012; Olmos and Albano 2012; Bezerra et al. 2013). As a consequence, fieldwork lasting for 40 or less days represented 41% out of all inventories examined in this review, while those involving 80 or more days of fieldwork corresponded to 6% of them (Suppl. Mat. Table S1, Fig. 4).

Fig. 4
figure 4

Number of days involved in the samplings of 32 avian inventories conducted in the Caatinga and published in the last three decades (1990–2019). The other 19 inventories examined in this review have not provided information on this aspect of the sampling effort

Full size image

Number of hours of field activities

When providing information on the number of hours spent in field activities, investigators separated the sampling effort into two categories, based on complementary procedures to record birds: 1) the number of hours spent with mist-netting, and 2) the number of hours involving bird detection through sight, vocalization, photographs and/or collects. These complementary procedures were adopted simultaneously in 31 (61%) of the 51 inventories (e.g. Araújo et al. 2012; Nunes and Machado 2012; Schunck et al. 2012), while the other 20 studies adopted only one of them (e.g. Farias 2009; Pereira and Azevedo 2011).

Among inventories of this latter group, 18 studies have not used mist-nets to sample communities (Suppl. Mat. Table S1), thus having recorded birds by sight, vocalizations, photos and/or collects (e.g. Parrini et al. 1999; Las-Casas et al. 2012; Bezerra et al. 2013). All of them properly informed the number of hours employed in fieldwork. On the other hand, only Lima et al. (2003) and Roos et al. (2006) sampled communities exclusively through mist-netting, and also informed the sampling effort (Suppl. Mat. Table S1). Thus, when only one of these complementary procedures was used (with or without mist-nets), the investigators informed the number of hours employed in field activities.

Sampling in rainy and dry periods

Nine (18%) studies have not informed whether field activities occurred during dry or rainy periods (Suppl. Mat. Table S1). On the other hand, most (n = 34; 67%) studies inventoried bird communities during both dry and rainy periods (e.g. Farias et al. 2005; Telino-Junior et al. 2005; Bezerra et al. 2013; Pichorim et al. 2014). Other six inventories investigated avifaunas only during rainy periods (e.g. Roos et al. 2006; Farias 2007; Ruiz-Esparza et al. 2011), while only two studies (Santos et al. 2012; Lima et al. 2019) sampled communities exclusively during the dry period. Thus, most inventories obtained data to provide information about seasonal variation on the species composition of communities. This acquisition of records in both dry and rainy periods throughout the Caatinga will help future studies to investigate the seasonal occupation of landscapes in response to the presence and absence of rainfall.

Evaluation and estimation of bird species richness

Most studies (n = 36; 71%) have not examined variation in the number of recorded bird species according to the sampling effort, nor produced a statistical estimation from the sampling data (Suppl. Mat. Table S1). These inventories occurred in the three decades involved in this review. They represent 100%, 79% and 66% of the studies published during the 1990s (Olmos 1993; Parrini et al. 1999), the 2000s (e.g. Nascimento et al. 2005; Farias 2007; Albano and Girão 2008) and the 2010s (e.g. Lima et al. 2011; Ruiz-Esparza et al. 2011; Barnett et al. 2014), respectively (Suppl. Mat. Table S1).

On the other hand, three studies simply observed the relationship between the number of species and the sampling effort, without the use of estimators. Among them, Olmos et al. (2005) examined variation in species richness according to the number of intervals of 30 min for eight localities in Ceará and Pernambuco, while two studies examined such variation in function of the number of expeditions to localities in Paraíba and Pernambuco (Telino-Junior et al. 2005; Pereira and Azevedo-Junior 2011).

Other 12 (23%) studies performed a statistical estimation of species richness from their samples. EstimateS 7.5 (Colwell 2005) was firstly used by Roos et al. (2006) and then by other five studies (e.g. Araújo and Rodrigues 2011; Lyra-Neves et al. 2012; Santos et al. 2012). Posterior versions of this Program were used by other six studies that also estimated species richness based on sampling data (e.g. Nunes and Machado 2002; Las-Casas et al. 2012; Mendes and Sousa 2016) (Suppl. Mat. Table S1).

These 12 studies employed four estimators in their analyses. Jackknife 1 was used in 100% of them (e.g. Araújo et al. 2012; Paixão 2012; Santos et al. 2012), while the use of Chao 2 was adopted in seven studies (e.g. Roos et al. 2006; Sousa et al. 2012; Mendes and Sousa 2016). Another estimator for incidence-based data (Bootstrap) was employed exclusively by Las-Casas et al. (2012), while three studies (Las-Casas et al. 2012, 2019; Brasileiro et al. 2017) used Chao 1, an estimator for abundance-based data (Colwell 2005; Gotelli and Colwell 2011) (Suppl. Mat. Table S1).

As a result, the statistical estimation of bird species richness from sample data has been used since 2006 by investigators that aimed to inventory Caatinga’s bird communities. Studies that performed such estimation represent only 29% of the 41 inventories published since then. Thus, most inventories published after Colwell (2005) do not contain this important information regarding sampling of wildlife communities.

Equipment used to record bird species

Binoculars and telescopes were the equipment used by investigators to sight birds at the study sites. In only five studies (e.g. Kirwan et al. 2001; Lima et al. 2003, 2011), observers used telescopes, especially in the first half of the 2000s, while binoculars were employed in all inventories (Suppl. Mat. Table S1, Fig. 5). Shotguns and a blowpipe were the equipment used by field workers to collect birds, without the need of capturing them alive, and were employed only during the first half of the 2010s. Shotguns were used by investigators in three studies (Silveira and Santos 2012; Vasconcelos et al. 2012; Santos et al. 2013), while Vasconcelos et al. (2012) was the only inventory in which specimens were collected with a blowpipe (Suppl. Mat. Table S1).

Fig. 5
figure 5

Temporal variation in the frequency of use of different equipment in avian inventories conducted in the Caatinga during the last three decades. The frequency represents the percentage of inventories that used a given equipment, considering the total number of inventories conducted during a five-year period

Full size image

Among the reviewed inventories, Lima et al. (2003) was the first study to use cameras to document species occurrence through photographs (Suppl. Mat. Table S1). After them, other researchers used cameras to photograph free birds (e.g. Olmos et al. 2005; Albano and Girão 2008; Pereira et al. 2014), and to document the recording of captured specimens (e.g. Freitas 2010; Schunck et al. 2012; Vasconcelos et al. 2012). In total, 20 (39%) studies used cameras to record birds in the Caatinga. Their use in inventories has been increasing since the early 2000s (Fig. 5).

Mist-nets were the only equipment used to capture birds, being employed in 27 (53%) studies (Suppl. Mat. Table S1). They were used in five of the six quinquennium involved in this review (Fig. 5). Olmos (1993) was the only study to use them during the 1990s. The use of mist-nets occurred with similar frequencies among inventories conducted during the last 20 years, being slightly more frequent during the first half of the 2010s (Fig. 5).

On the other hand, the combination of a sound recorder and a microphone was the procedure used to record and/or playback vocalizations of birds in the field (Suppl. Mat. Table S1). This pair of acoustic equipment was adopted in most inventories (n = 36; 71%). It was used by all inventories published in the 1990s, and it has been more frequent in inventories since the early 2000s. All inventories of the last quinquennium employed this equipment (Fig. 5).

Thus, the use of this range of equipment mentioned above led scientists to record birds through: 1) sight, 2) sound recording of vocalizations, 3) capture, 4) photographs, and 5) collecting. The recording of birds through the use of only sighting equipment occurred in five studies (e.g. Santos 2008; Araújo and Rodrigues 2011; Las-Casas et al. 2012). On the other hand, 17 (33%) studies involved two ways of recording specimens in the landscapes. Among them were ten studies that involved only the sighting and capture of specimens (e.g. Nascimento et al. 2000; Telino-Junior et al. 2005; Silveira and Machado 2012). The other seven studies involved equipment to sight or sound record birds (e.g. Parrini et al. 1999; Farias 2007; Kirwan et al. 2001).

A set of 19 (37%) studies used gear to obtain bird records through three ways. Freitas (2010) was the only study that used equipment to sight, capture, and photograph birds in the field. On the other hand, nine studies involved equipment to sight and capture birds, besides sound recording vocalizations (e.g. Roos et al. 2006; Araújo et al. 2012; Barnett et al. 2014). The other nine inventories recorded specimens through sights, sound recordings, and photographs (e.g. Olmos and Albano 2012; Bezerra et al. 2013; Pereira et al. 2014).

Seven studies recorded bird specimens through four ways. In six inventories, birds were recorded through sights, their capture, photographs, and vocalizations (e.g. Schunck et al. 2012; Silva et al. 2012; Sousa et al. 2012), while Silveira and Santos (2012) conducted their inventory by sighting, capturing, sound recording and collecting birds with shotguns. Further, Vasconcelos et al. (2102) was the only study to record birds through five ways - sight, sound recording, capture, photographs, and collecting with shotguns and a blowpipe.

Techniques used to sample bird communities

Roda and Carlos (2004) was the only study to do not inform on the techniques employed to sample the bird communities (Suppl. Mat. Table S1). The detection of bird species through random walks along pre-existing trails and/or roads was the technique most often employed in the 50 inventories that provided information on this aspect of the methods (n = 49; 98%). This technique was nearly used in all studies of the six quinquennium involved in this review (Fig. 6). The capture of birds with mist-nets occurred in more than half of the studies that informed about techniques (n = 30; 60%). Mist-netting occurred in a single inventory, published in the first half of the 1990s (Olmos 1993), and were not employed in the following five years. Mist-nets were used in 30–71% of the inventories in the four more recent quinquennium, with no clear tendency to increase or decrease. Fourteen studies involved collecting, with Parrini et al. (1999) being the only study to use this technique in the 1990s. The other thirteen inventories that collected birds were restricted to the first half of the 2010s, representing 46% of the studies conducted in this period (Fig. 6, Suppl. Mat. Table S1).

Fig. 6
figure 6

Temporal variation in the frequency of use of different techniques in inventories of bird species conducted in the Caatinga during the last three decades. The frequency represents the percentage of inventories that adopted a given technique to sample the communities, considering the total number of inventories conducted during a five-year period

Full size image

Lesser common techniques were walks with stops at regular intervals to count individuals (sensu Willis 1979). They were employed in six (12%) inventories, being more frequent among studies published in the 2005–2009 than in the last decade (Fig. 6). The elaboration of lists of 10 or 20 species (sensu Mackinnon et al. 1993) occurred in only five publications of the 2010–2014 quinquennium. Surveys through point counts (sensu Bibby et al. 1992) were adopted in seven inventories, and its use appear to be increasing in frequency among studies published in the last 15 years (Suppl. Mat. Table S1, Fig. 6). Sousa et al. (2012) was the only study to use transect counts to sampled bird communities (Suppl. Mat. Table S1).

Point counts and random walks were the only techniques to be employed alone, as observed in one and 11 studies, respectively (e.g. Farias et al. 2010; Bezerra et al. 2013; Pereira et al. 2014; Mariano 2015, 2016). On the other hand, some authors combined the use of two techniques during their sampling effort in an attempt to obtain records of more species. For example, the combination of random walks and captures with mist-nets was adopted in 19 (37%) of the reviewed studies (Suppl. Mat. Table S1). The use of three or more techniques to sample communities initiated in inventories published in the early 2010s (e.g. Dornellas et al. 2012; Nunes and Machado 2012; Silveira and Santos 2012; Vasconcelos et al. 2012). These inventories combined techniques such as the elaboration of Mackinnon Lists, point counts, and collects, and comprised 22% (n = 11) of the inventories examined in this review (Suppl. Mat. Table S1).

Gathering quantitative data

Eleven (22%) studies have used techniques targeting quantitative information about the species occurrence. Five of them have shown data on the relative abundance of species by providing the numbers of birds recorded per 100 h of observation. Four of these studies (Olmos et al. 2005; Farias 2007; Albano and Girão 2008; Farias 2009) followed procedures performed by Willis (1979), while Santos et al. (2012) cited Willis and Oniki (1981). On other hand, quantitative data obtained from captures with mist-nets was found in four studies (Lyra-Neves et al. 2012; Paixão 2012; Sousa et al. 2012; Freitas et al. 2016). Less frequent were studies that obtained quantitative data through the employment of point counts (Araújo and Rodrigues 2011; Las-Casas et al. 2019), transect counts (Sousa et al. 2012), and the elaboration of Mackinnon Lists (Nunes and Machado 2012).

Destination of bird skins

A total of 14 studies involved the collect of bird specimens during their activities in the study sites. These birds were collected through captures with mist-nets, and the use of shotguns and blowpipe (Suppl. Mat. Table S1). Collect was the main intention of investigators in some studies, while other collects occurred because a range of birds died during their capture with mist-nets (Suppl. Mat. Table S2).

Bird specimens collected in the Caatinga during inventories published in the last 30 years were deposited in ornithological collections of eleven institutions (Suppl. Mat. Table S2). Museu Paraense Emílio Goeldi was the only institution located in the northern geopolitical region of Brazil. Other four and six institutions that received specimens are located in southeastern and northeastern Brazil, respectively. Universidade Estadual de Feira de Santana, in Feira de Santana, and Universidade Federal de Pernambuco, in Recife, were the institutions that received birds collected by higher numbers of studies. Birds were collected in five states of eastern Brazil, of which Bahia received more inventories involving this type of activity (Suppl. Mat. Table S2).

It is important to highlight that my review considered only published inventories, thus not encompassing collects that have not been reported by collectors, curators of collections, or other scientists. Collecting has been important for the generation of knowledge about the Caatinga’s birds since previous centuries (Pacheco 2002; Silva et al. 2003; Araújo and Silva 2017), and certainly will keep welcome in future inventories.

Sound and photographic archives that received digital vouchers

Digital vouchers are valuable in avian inventories for providing documentary evidence of species, and include sound recordings and photographs (Carlos et al. 2010; Lees et al. 2014). Sound recordings of Caatinga’s bird species obtained by eight inventories were deposited in two overseas and six Brazilian archives. Arquivo Sonoro Professor Elias Coelho, Universidade Federal do Rio de Janeiro and the Wikiaves database received vouchers from a higher number of studies. Bahia was the state most often sampled by these studies (Suppl. Mat. Table S3). In an attempt of providing further documentation of the recorded species, some authors archived digital vouchers in two Brazilian photographic databases (Suppl. Mat. Table S3). WikiAves received photographs taken in Caatinga portions located in Ceará, Bahia and Pernambuco (Vasconcelos et al. 2012; Barnett et al. 2014; Brasileiro et al. 2017; Las-Casas et al. 2019). Further, photographs made by Albano and Girão (2008) in three mountain ranges of Ceará were archived in the Aquasis’ database.

The documentation of species occurrence through sound recordings and photographs complements that promoted by collects (Carlos et al. 2010; Less et al. 2014). According to them, the obtention of these digital vouchers is desirable for as many species as possible in every avian inventory. For the Caatinga, the obtention of digital vouchers became more frequent in inventories published during the 2010s, and is likely to keep occurring in future inventories. This is because the continuous development of technologies regarding optical and bio-acoustic equipment will likely increase the ease in their obtention.

Conclusion

Factors such as logistics, financial support, time availability and personal goals might have contributed to the observed great variation in the length of sampling periods.

Also, the Caatinga’s harsh environment obviously leads to difficulties in the field even for hard field workers.

The participation of few or several investigators in a given inventory appears to be leading to distinct situations (Suppl. Mat. Table S1). First, inventories conducted by several contributors tended to sample communities for longer periods than those that had only 1–2 field workers. Second, field activities involved larger sets of equipment and techniques when done by several than by a few field workers. Third, publications with several authors tended to present more complete information regarding the sampling effort than those elaborated by 1–2 authors. The lack of information relative to the number of hours spent in fieldwork was more frequent than those regarding the numbers of days, months, and years (Suppl. Mat. Table S3).

The participation of several field workers in a given inventory might also contribute to the reduction of the number of species identification errors, that might occur eventually in avian inventories (Lees et al. 2014). This is because the knowledge and attention associated with several ornithologists is more likely to lead to the production of error-free inventories than those associated with 1–2 observers. Cases of misidentification can proliferate in the literature unless they are quickly recognized (Lees et al. 2014). For example, Silva et al. (2003) have not included in their checklist of the Caatinga’s avifauna six species (Harpagus bidentatus, Glaucis hirsuta, Thalurania watertonii, Automolus leucophthalmus, Thryothorus leucotis and Cyanerpes cyaneus) recorded in the inventory conducted by Nascimento et al. (2000) at Chapada do Araripe. Thus, it would be important for authors of inventories to verify compatibility of the known geographic distribution of the species with the location of the new record, mainly in cases of species not supported by documentary evidence. The publication of unmistakable records has been a concern of members of the Brazilian Ornithological Records Committee (CBRO) for several years (Piacentini et al. 2015). This should also occur for all avian inventories conducted in the Caatinga and other biomes. Regarding this concern, a set of suggestions targeting the publication of accurate and documented records has been provided by Lees et al. (2014).

Suggestions for further inventories in the Caatinga

A range of suggestions regarding documentation of ornithological records and avian inventories in Brazil has been provided by Carlos et al. (2010) and Lees et al. (2014). To avoid been repetitive, I focus on aspects not addressed by them. My suggestions for future avian inventories in the Caatinga are:

  • Ornithologists could try to provide information on all aspects regarding the field activities and the sampling effort. Thus, authors should worry in presenting enough information for replication of their studies.

  • Investigators could try to spread their sampling effort throughout the several periods of a given year. This means having field activities in rainy and dry periods, and along the four seasons, thus being able to detect migratory species coming from other biomes.

  • Authors of publications could provide a map of the study area showing the localities where field activities have been done. This is especially important for protected areas, as some are huge and might encompass variable bird communities within their limits. Thus, sampling sites would lead to a better inventory if distributed throughout the whole protected area.

  • Authors could produce a statistical estimation of bird species richness from the sample data (sensu Colwell 2005). This would lead to the elaboration of checklists containing information regarding the observed and expected species richness, thus bringing information on the adequacy of the sampling (see Magurran 2004; Gotelli and Colwell 2011). Thus, this permits to conclude if further sampling effort would be necessary to know the real bird species richness of a given area in the Caatinga. These estimations would be important for future studies targeting the production of local and regional checklists, and comparisons of bird faunas among several areas.

  • Besides concerning documentation (Carlos et al. 2010; Lees et al. 2014), investigators could try to produce quantitative data for as many species as possible, thus complementing the essential sampling through random walks, that permits the investigators to obtain information along extensive areas. Complementary techniques that provide quantitative data about species tend to sample communities in restricted areas but certainly will produce important additional information for bird conservation through the Caatinga. This would be possible by adopting some of the numerous techniques that are available in the literature. Although several techniques have been employed in the Caatinga (Suppl. Mat. Table S3), some have not been used in this dry forest region. For example, the standardized search proposed by Watson (2003) has not been frequently adopted in Brazil, but could be employed in the Caatinga. His approach allows the avifauna of entire sites (portions of landscapes) to be sampled and compared directly, producing comparable measures of species richness, and estimates of sample completeness and species incidence.

  • As inventories deal with entire communities, their improvement can be achieved with the use of multi-species occupation modeling (MSOM). These models have been considered as a promising and efficient approach for the study of communities worldwide (Devarajan et al. 2020). This is because they can improve the quality of inventories by incorporating imperfect detection of species – what occurs when a species was not detected due to reasons such as insufficient sampling, and aspects regarding species detectability. According to them, MSOM can jointly estimate detection and occurrence probabilities of each species in a community, including those not detected by field workers. These models are typically used to estimate species richness at the community level, and to make inferences about community-level patterns of occupancy. Devarajan et al. (2020) elaborated a roadmap for the use of MSOM in community studies that could be followed in future avian inventories conducted in the Caatinga.

  • Further, autonomous recording units are a widely used sampling tool in ecological research and monitoring (Shonfield and Bayne 2017). According to their important review, this type of equipment can record avian sounds in a wide range of habitats, including remote locations and harsh environments. Therefore, future inventories of bird species in the Caatinga could consider using a set of these recording units during their samplings. Recommendations relative to the use of this type of sound recording equipment/technique also have been recently proposed by Darras et al. (2019). In analogy, camera-traps could be used as complementary equipment to fulfill some gaps in future inventories done in the Caatinga. Although they can record a wide diversity of species (e.g. Cloyed et al. 2018), they would be valuable in obtaining important data especially of relatively large and/or less common birds of a range of families, including Picidae, Cuculidae, Cracidae, Tinamidae, and even the Greater Rhea, Rhea americana, in this Brazilian region. O'Brien and Kinnaird (2008) reviewed the use of camera-traps in ornithological studies, and thus their study could be consulted by researchers interested in conducting further avian inventories in the Caatinga.