Journal of Arid Environments 74 (2010) 344–349

Contents lists available at ScienceDirect

Journal of Arid Environments

journal homepage: www.elsevier.com/locate/jaridenv

Biological nitrogen fixation in tree legumes of the Brazilian semi-arid

A.D.S. Freitas a,*, E.V.S.B. Sampaio b, C.E.R.S. Santos a, A.R. Fernandes b a Universidade Federal Rural de , Departamento de Agronomia, Rua Dom Manoel de Medeiros, s/n, (PE), 52171-900, b Universidade Federal de Pernambuco, Departamento de Energia Nuclear, Avenida Prof. Luı´s Freire, 1000, Recife (PE), 50740-540, Brazil article info abstract

Article history: Information on biological nitrogen fixation (BNF) in dry forests is scarce, in spite of the importance of the Received 2 July 2008 process to nitrogen cycling in these systems. It can be used to improve management and to select species Received in revised form for revegetation and agroforestry systems. The BNF of tree legumes was estimated in four areas of the 12 March 2009 Brazilian dry forest (caatinga), analyzing potentially fixing species (target species) and non fixing species Accepted 15 September 2009 (reference species), using the 15N natural abundance methodology. The differences between the isotopic Available online 7 October 2009 signals of target and reference species permit the safe identification of N2-fixing plants and reasonably precise calculations of proportions of N derived from the air (%Ndfa). Species with great N fixation capacity Keywords: 2 Dry forest were identified, notably Mimosa tenuiflora, Mimosa arenosa and Piptadenia stipulacea. The mean biological Natural abundance fixation contributions to the plants were high, varying from 27 to 68%. However, the estimated quantities of 1 1 Northeast Brazil N added annually to the leaf biomass in the systems were low (2.5 and 11.2 kg ha year ), due to low N-15 proportions of fixing plants in the vegetation communities. In situations of native vegetation regeneration, when the succession is dominated by fixing species, fixation could reach up to 130 kg ha1 year1. Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction Tropical dry forests are usually rich in legume species and grow in soils with low organic N; therefore, the contribution of biological Biological fixation is the principal process of nitrogen entry into fixation to these species could be high (Shearer et al., 1983; Ndiaye natural ecosystems and the fixation associated with vascular plants and Ganry, 1997; Raddad et al., 2005). However, not all legume usually contributes with the greatest quantities of added N species are capable of fixation and, among the fixing ones, the (Cleveland et al., 1999). However, estimates of these contributions proportion of fixed N varies greatly (Allen and Allen, 1981; Faria are scarce because the traditional methods of measuring fixation do et al., 1984, 1987, 1994; Faria and Lima, 1998; Boddey et al., 2000). not work in trees and for extended periods of time, except for the Even nodulated trees of the same species vary in fixation, with method based on variations of the 15N natural abundance. Even this a tendency of lower proportions in old growth vegetation than in method only allows reasonable estimates if there is a large differ- regrowth (Gehring et al., 2005). ence between the 15N abundance of fixing and non fixing plants, Brazil has one of the largest semi-arid areas in the world (about a condition not found in many ecosystems (Handley et al., 1994; 1 million km2), half of which covered with a native dry forest, called Roggy et al., 1999; Gehring and Vlek, 2004). Where the 15N condi- caatinga, with great richness and abundance of legume species tions are appropriate, most of the data refer only to the proportion (Queiroz, 2006). The nodulation capacity of most of these species is of fixed N in relation to the total N of single trees, without quan- unknown and no one determined the proportion of N derived from tifying the masses of fixed N in these trees (Boddey et al., 2000). As fixation in plants growing in this area. A previous work (Freitas, a consequence, the amounts added by N fixation of vascular plants 2008) showed that non fixing nitrogen caatinga plants are very to the ecosystems were estimated in very few cases and most of much enriched with 15N and that this enrichment is uniform, as them dealing with temperate forests (Cleveland et al., 1999). much spatially as between species. This isotopic standard is Among the rare ones dealing with tropical forests most referred to a promising condition for use of the 15N natural abundance meth- humid sites (Roggy et al.,1999; Gehring et al., 2005) and no one was odology for BNF estimates, because it facilitates the choice of found dealing with tropical dry forests. reference plants, the sampling scheme and the detection of significant differences between the d15N of fixing and non-fixing plants (Ho¨gberg, 1997; Boddey et al., 2000). The estimation of biological nitrogen fixation (BNF) in plants of * Correspondence author. R. Ja´der de Andrade, 404/1801, Recife (PE), 52061-060, Brazil. Tel.: þ55 81 21267979. this area would increase the knowledge of fixation in native vege- E-mail address: [email protected] (A.D.S. Freitas). tation of semi-arid areas. Identifying high fixing species adapted to

0140-1963/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.jaridenv.2009.09.018 A.D.S. Freitas et al. / Journal of Arid Environments 74 (2010) 344–349 345 these areas may contribute to regevetation efforts, better manage- Table 2 ment of native vegetation as pastures or fallow in shifting cultivation Characteristics of the soils under caatinga vegetation in four chosen municipalities in the States of Paraı´ba (PB) and Pernambuco (PE), Brazil. and selection of trees to be used in agroforestry systems. Moreover, a few places have data on legume proportion and on above ground Characteristics Municipality (state) biomass production which could be used to estimate the quantities Remı´gio Santa Teresinha Serra Talhada of N fixed in the vegetation. Considering that, apparently, this (PB) (PB) (PE) (PE) information is not available for any native vegetation in tropical Soil type Regolithic Litholic Yellow Luvisol semi-arid areas, it would provide the first data on the importance of Neosol Neosol Argisol fixation in similar areas, allowing comparisons with those of humid pH (water) 4.4 5.8 5.2 6.8 P (mg dm3) 8.4 7.3 2.8 4.9 tropical (Roggy et al.,1999; Gehring et al., 2005) and temperate areas Sand (g kg1) 725 623 654 651 (Cleveland et al., 1999). Silt (g kg1) 122 224 223 227 Thus, the objectives of this work were to determine BNF in tree Clay (g kg1) 153 153 123 122 legumes of caatinga, using the 15N natural abundance method- ology, to identify species with great fixing potential and to make the first estimates of the quantities of N added annually to the leaf nodulation capacity or without information on nodulation (Allen biomass of this vegetation. We hypothesized that: (1) the 15N and Allen, 1981; Faria et al., 1984, 1987, 1994, Faria and Lima, 1998). natural abundances of fixing legume trees are significantly lower Samples were collected from all the target species found in each plot than those of non fixing species, allowing good estimates of their and also from four reference species (Table 3). These reference BNF; and (2) BNF represents large proportions of total leaf nitrogen species were two non legume species (Croton sonderianus Muell. in some plants and species. Arg. and Aspidosperma pyrifolium Mart.) and two non nodulating legume species (Bauhinia cheilanta (Bong.) Steud and Caesalpinia 2. Material and methods pyramidalis Tul.). They were selected because of their wide distri- bution and high density in the caatinga vegetation areas. Due to the 2.1. Studied areas absence of plants of some species in some plots, the number of plants of the species were, in most cases, lower than the number of Four fragments of native caatinga were selected in four plots in each site. municipalities of the semi-arid regions of the states of (PB) Leaf material from different branches was collected from adult and Pernambuco (PE). Remı´gio (PB) and Caruaru (PE) are localized plants with diameters at breast height larger than 3 cm. After in the Agreste physiographic zone and Santa Teresinha (PB) and collection, completely expanded healthy green leaves were Serra Talhada (PE) are localized in the Serta˜o zone, and these zones selected, forming a composite sample of around 100 to 200 g of reflect the different water availability conditions for plants (inten- leaves per plant. In all the areas the samples were collected during sity and duration of water deficit period) in the fragments studied the rainy season. From all the sample species, botanical material (Table 1). was collected for comparison with the collections deposited in Descriptions of flora composition and vegetation structure were herbarium, of floristic surveys carried out by Alcoforado Filho et al. published for the forest fragments of Caruaru (Alcoforado Filho (2003), in Caruaru, Ferraz et al. (2003), in Serra Talhada, and Pereira et al., 2003), Serra Talhada (Ferraz et al., 2003) and Remı´gio (Pereira et al. (2002), in Remı´gio. et al., 2002). Number of species, height and diameter of plants are greater in the Agreste sites than in Serra Talhada. There is no published description of the caatinga fragment of Santa Teresinha, 2.3. Determination and statistical analysis but number of species, height and diameter of plants are slightly lower than those of Serra Talhada, probably reflecting a lower water The leaf material was oven dried at 60 C for 72 hours and then availability. Characteristics of the soils of the four fragments studied ground to fine powder. Total N was analyzed by micro Kjeldhal are presented in Table 2. methodology and 15N abundance by mass spectrometry and expressed in ‘‘delta’’ notation, which is the deviation per thousand 15 2.2. Sampling scheme (&)of N abundance of the sample in relation to the standard:   Five plots were established in Remı´gio and Serra Talhada and six d ¼ Rsample=Rstandard 1 1000 plots in Santa Teresinha and Caruaru. Each plot had a 20 m radius 15 14 and was 50 m or more distant from any other. Species potentially where Rsample and Rstandard are the ratio N: N of the sample and the standard (atmospheric N2), respectively. able to fixate N2 (target species), or not (reference species) were sampled in each plot. The target species were legumes with a known The data were submitted to a statistical analysis to compare the mean isotopic signals of the target legume species with the refer- ence species signals, in each site, using the values of each individual Table 1 of the same species as replications. Significant differences were General characteristics of the four chosen municipalities, in the States of Paraı´ba (PB) established considering the 0.01 level of probability of the Tukey and Pernambuco (PE), Brazil. test. The species whose mean were significantly impoverished in Characteristics Municipality (state) 15N in relation to the reference species were considered as being Remı´gio Santa Teresinha Caruaru Serra Talhada capable of atmospheric N2 fixation (fixing species). (PB) (PB) (PE) (PE) Coordinates 652’S 0703’S 0815’S 0759’S 3547’W 3729’W 3557’W 3818’W 2.4. Estimation of BNF Altitude (m) 596 380 545 500 Annual rainfall (mm) 700 824 696 768 For the N2-fixing species, estimates of percentage of plant Months with water deficit 4–5 9–10 4–5 6–7 nitrogen derived from the air (%Ndfa) were made, using the Average temperature (C) 22 26 24 24 formula recommended by Shearer and Kohl (1986): 346 A.D.S. Freitas et al. / Journal of Arid Environments 74 (2010) 344–349

Table 3 annual leaf production estimated by Dantas (2003). For the Serra 15 Leaf d N, nodulation and nitrogen derived from the atmosphere (%Ndfa) in plants Talhada fragment, the estimates were carried out using basal area collected at four different sites in the semi-arid caatinga of Paraı´ba and Pernambuco determinations done by Ferraz et al. (2003), and the leaf mass States, Brazil. measurements made by Kauffman et al. (1993). Species na d15N& Nodulationb %Ndfac

(average standard (I) (II) 3. Results deviation)

Santa Teresinha 24 36 27 3.1. Identification of N -fixing species Reference plantsd 12 6,71 1,14 2 e Piptadenia stipulacea 5 4,43 1,68 ** w.i. 34 26 Mimosa tenuiflora 5 4,22 0,91 ** nod 37 28 In the four sites studied, species of legumes with isotopic signals Anadenanthera colubrina 2 6,73 0,78 ns nod – – (d15N) similar to the mean of the reference species (Bauhinia chei- Serra Talhada 45 68 52 lanta, Caesalpinia pyramidalis, Aspidosperma pyrifolium and Com- Reference plantsd 20 6,58 1,41 ns miphora leptophloeos), as well as species with significantly inferior Parapiptadenia zehntneri 4 4,00 1,11 ns w.i. - - Anadenanthera colubrina 4 5,27 1,62 ns nod - - signals were identified (Table 3). The species with statistically Mimosa tenuiflora 4 1,59 1,06 ** nod 76 58 lower signals were considered as species that were able to obtain Mimosa sp1 4 2,05 0,64 ** w.i. 69 53 part of their N supply through biological fixation of atmospheric N2. Mimosa sp2 5 2,39 0,32 ** w.i. 64 49 All the plants of Pithecellobium parviflorum Benth, Piptadenia stip- Mimosa sp3 4 2,75 1,05 ** w.i. 63 48 Remı´gio 42 49 41 ulacea (Benth.) Ducke and Mimosa spp (M. arenosa (Willd.) Poir., M. Reference plantsd 19 9,84 0,65 tenuiflora (Willd.) Poir., M. paraibana Barneby and five other non Mimosa paraibana 1 4,50 ** w.i. 54 45 identified species) presented 15N impoverished leaves in relation to Mimosa arenosa 5 5,59 0,96 ** nod 43 36 reference species. The plants of Anadenanthera columbrina (Vell.) Mimosa sp4 3 5,38 2,57 ** w.i. 45 38 Brenan collected in Caruaru were considered as fixing, but those Mimosa sp5 2 4,18 1,70 ** w.i. 58 48 Mimosa tenuiflora 3 3,47 1,02 ** nod 65 54 collected in Santa Teresinha and Serra Talhada had isotopic signals Piptadenia viridiflora 5 9,99 2,31 ns nod - - similar to those of the reference species. Piptadenia zehntneri Piptadenia stipulacea 4 7,09 0,78 ** w.i. 28 23 (Harms) M.P. de Lima & Lima and Acacia paniculata Willd. had Caruaru 34 64 52 isotopic signals equal to those of the control. Reference plantsd 10 9,03 1,81 Anadenanthera colubrina 5 5,34 2,01 ** nod 41 33 Pithecellobium parviflorum 4 5,28 0,93 ** w.i. 42 34 3.2. BNF estimate Acacia paniculata 4 10,39 1,07 ns w.i. - - Mimosa arenosa 6 1,04 0,74 ** nod 88 72 Piptadenia stipulacea 5 1,28 0,47 ** w.i. 86 70 Species with great N fixation capacity (Ndfa > 50%) were found in Serra Talhada, Remı´gio and Caruaru, even considering the most a Number of plants. b & Literature information on nodulation occurrence: w.i.. ¼ without information; conservative estimates (%Ndfa calculated considering B ¼2 ). nod ¼ nodulating. The most remarkable ones were Mimosa tenuiflora, in Serra Talhada c %Ndfa calculated with B value of 0 (I) or –2 & (II). (58% of Ndfa) and in Remı´gio (54% of Ndfa), and Mimosa arenosa d Croton sonderianus, Aspidosperma pyrifolium (non legume species), Bauhinia (72% of Ndfa) and Piptadenia stipulacea (70% of Ndfa), in Caruaru. cheilanta and Caesalpinia pyramidalis (non nodulating legume species). The mean biological fixation contribution for plant nitrogen was e Averages significantly different (**) or not (ns) of reference plants averages, at 0,01 probability level (Tukey test). 27% or 36%, in Santa Teresinha, 41% or 49%, in Remı´gio, 52% or 64%, in Caruaru, and 52% or¨68%, in Serra Talhada, according to calcu- lations that were done considering B ¼2orB¼ 0&, respectively   (Table 3). 15 15 15 The lowest N content of leaves (2.01%) was found in the non- %Ndfa ¼ d NðreferenceÞ d NðN-fixingÞ=d NðreferenceÞ B 100 legume reference species (Aspidosperma pyrifolium and Commi- 15 15 phora leptophloeos)(Table 4). Among the legumes, the N leaf content Where d N(reference) is the mean value of the d N of the reference 15 15 differed significantly (p 0.01) between the fixing and the non- species of each site, d N(N-fixing) is the mean d N value for the 15 fixing groups of species. The fixing legumes had a mean content of plants of each N2-fixing species and B is the d N value for fixing plants cultivated in the absence of soil mineral N supply. Due to the 15 high d N values found for non fixing plants of the caatinga and Table 4 methodological complications for estimating this value in arboreal Average leaf nitrogen concentrations of plants collected at four caatinga sites of species (Ho¨gberg, 1997; Boddey et al., 2000), the B values in this Paraı´ba and Pernambuco States, Brazil. Numbers in parentheses refer to replicates. work were not estimated. However, according to the suggestion by Sites Fixing Non-fixing Non Overall Ho¨gberg (1997), the importance of using extreme B values in %Ndfa legumesa legumesb legumes averages calculations was tested. With the absence of data for the studied Remı´gio 3.90 0.51 2.87 0.30 2.21 0.36 3.13 0.86a species, values of 0& and 2& were used which are commonly (18)c (13) (11) found in studies of tree legumes.(Schulze et al., 1991; Roggy Caruaru 2.95 0.54 2.24 0.50 1.81 0.20 2.60 0.66b (20) (9) (5) et al.,1999; Raddad et al., 2005). Serra Talhada 2.88 0.74 2.26 0.44 1.78 0.27 2.38 0.69bc 1 1 The quantities of fixed N (kg ha year ) were estimated using (17) (18) (10) %Ndfa and %N content results, obtained in this work, and the annual Santa Teresinha 2.19 0.35 1.80 0.36 1.60 0.15 1.93 0.40c production of leaf mass of the fixing species, estimated from (10) (10) (4) information in other available works only for the Remı´gio and Serra Averages 3.10 0.83 A 2.48 0.69 B 2.01 0.37 C Talhada fragments. Due to the low number of replications per a Target species with d15N significantly lower than the mean of reference plants. b 15 species the mean %Ndfa and %N of all the fixing species for each site Bauhinia cheilanta, Caesalpinia pyramidalis and target species with d N similar to the mean of the reference species. were used. For the Remı´gio fragment the primary liquid production c Number of plants (replicates) Averages followed by the same capital letter in the of the fixing legumes was estimated based on the relative basal area line and small letter in the column are not significantly different at the 0.01 level of these species, determined by Pereira et al. (2002), and on the (Tukey test). A.D.S. Freitas et al. / Journal of Arid Environments 74 (2010) 344–349 347

Table 5 already been observed in two species of Parapiptadenia (Allen and Total biomass, total nitrogen and fixed nitrogen in leaves of fixing legumes at two Allen, 1981; Faria et al., 1984), an indigenous genus of the Brazilian caatinga sites (Serra Talhada and Remı´gio) of semi-arid Northeast Brazil. Northeast and very close to the nodulating genera Piptadenia and Variable Serra Talhada Remı´gio Anadenanthera (Allen e Allen, 1981). Therefore, it is probable that Total leaf biomass (ton ha1) 6.6a 5.3b the high 15N signal of Parapiptadenia zehntneri plants is due to some Proportion of fixing legume plants (%) 2.4c 11.8d physiological and/or environmental restriction for the N fixation 1 2 Biomass of fixing plants (kg ha ) 170 625 and not to lack of nodulating capacity. On the contrary, Acacia N content in fixing legume leaves (%) 2.88 3.90 Stock of N in fixing plants (kg ha1) 4.9 24.4 paniculata is probably not a N2-fixing species, since the majority of %Ndfa, with B ¼ 06849Acacia species of the Americas, in contrast with the African ones, do %Ndfa, with B ¼25241not form symbiosis with rhizobia (Allen and Allen, 1981; Faria et al., 1 Estimate of fixed N, with B ¼ 0 (kg ha ) 3.3 11.2 1984, 1994; Faria and Lima, 1998). Direct observation studies of Estimate of fixed N, with B ¼ –2 (kg ha1) 2.5 9.3 radical nodules need to be done for definitive identification of the a Kauffman et al. (1993). nodulation capacity or these species. b Dantas (2003). c Ferraz et al (2003). d Pereira et al (2002). 4.2. Estimations of BNF

Results of isotopic signals provided very promising conditions 3.10%, while the non fixing had 2.48% of N in the dry biomass of the for calculating N derived from biological fixation. The signals of the leaves. The highest N content was found in the Remı´gio trees (3.23%) reference plants were high in all the sites studied (Table 3) and in while the lowest was found in Santa Teresinha (1.93%). the fixing plants they were more than two units of d15N& lower Utilizing the available total biomass leaf data (Dantas, 2003) and than those of the reference plants. The differences varied between proportion of fixing species (Pereira et al., 2002), for Remı´gio, the 2.28& (Piptadenia stipulacea, in Santa Teresinha) and 7.98& annual production of leaves of fixing species was estimated at (Mimosa arenosa, in Caruaru). This condition, that permits safety in 625 kg ha1 (Table 5). A similar estimation, for Serra Talhada, using fixation studies (Ho¨gberg, 1997), is not often found, principally published data of leaves mass (Kauffman et al., 1993) and fixing in humid forests (Roggy et al., 1999; Gehring and Vlek, 2004). Even species proportion (Ferraz et al., 2003), resulted in 170 kg ha1. in semi-arid environments the differences between the d15N signals From these results and averages, per site, of N content and %Ndfa, of potentially fixing species and non fixing species can be ambig- the quantities of N added annually to the leaf biomass in the systems uous, generating doubts as to whether the fixing species are fixing were estimated. In Serra Talhada, the total quantity of N fixed from a little or if the method does not work (Handley et al., 1994). As the atmosphere by the tree legumes would be 3.3 kg ha1 year1, a consequence, there are very few studies with clear evidence of considering B ¼ 0, or 2.5 kg ha1 year1, considering B ¼2(Table 5). fixation in native vegetations of dry ecosystems in the world In Remı´gio, these values varied between 11.2 kg ha1 year1, (Shearer et al., 1983; Ndiaye and Ganry, 1997). adopting B ¼ 0, and 9.3 kg ha1 year1, adopting B ¼ –2. High contributions of BNF were estimated for the tree legumes of the caatinga, which had mean %Ndfa, in the minimum, of 27% in 4. Discussion Santa Teresinha, and, in the maximum, 52% in Caruaru and Serra Talhada, considering B ¼ –2& (Table 3). If the calculations had been 4.1. Identification of N2-fixing species done with B ¼ 0, these values would have increased to 36 and 68%, respectively. Comparison of these results with those found in Among the 14 target species, 8 species of Mimosa spp (M. are- literature, as much for site averages as for the species, were done nosa, M. tenuiflora, M. paraibana and another five non identified utilizing B ¼ –2&, a value that is commonly used in tree legume species), Pithecellobium parviflorum, Anadenanthera columbrina and studies (Shearer et al., 1983; Schulze et al., 1991; Sprent et al., 1996; Piptadenia stipulacea presented (d15N) isotopic signals significantly Ndiaye and Ganry, 1997; Roggy et al. 1999; Raddad et al., 2005). inferior to the reference species, being identified as fixing species. However, in this work, the impact of using B ¼ 0orB¼ –2& in the Although there are no previous reports on the nodulation capacity estimates of %Ndfa was small (Table 3). of M. paraibana, P. parviflorum and P. stipulacea, these results are Although higher values of %Ndfa are found in some highly effi- consistent with the nodulation capacity information on other cient species in symbiotic nitrogen fixation, as much under natural species of these genera (Allen and Allen, 1981; Faria et al., 1984; conditions (Sprent et al., 1996) as in plantations (Gathumbi et al., 1994; Faria and Lima, 1998). Anadenanthera columbrina (a nodulat- 2002), the mean values for sites with natural vegetation are generally ing species, according to Faria et al.,1984) varied in signal among the much lower than those found for the caatinga. Ndiaye and Ganry sites, in Caruaru being considered as fixer and in Serra Talhada and (1997) found very small contributions (Ndfa ¼ 10% and 20%) of BNF Remı´gio as non fixer. Piptadenia viridiflora, a nodulating species in two ecological zones localized in semi-arid regions of Senegal, (Faria and Lima,1998), presented a non fixing signal. The behavior of while Schulze et al.(1991) found a mean of 32% of BNF contribution in these Mimosoideae points to the fact that the occurrence of N2 an arid gradient in Namibia. Similar values (43 to 65%) to those found atmospheric fixation in legumes with nodulating capacity is influ- in Serra Talhada and the Agreste fragments were found in some enced by environmental and/or physiological conditions. In systems desertic areas of California covered by Prosopis spp.(Shearer et al., where the N stock is well conserved, fixing legumes may obtain 1983). However, it is relevant to say that such plants, in spite of being small quantities of BNF (Boddey et al., 2000). In mature forests, large in a desertic environment, have a phreatrophic habit, growing under size plants can present physiological limitations to BNF (Gehring more favorable conditions for BNF. In a humid forest in French et al., 2005). Guyana, Roggy et al.(1999) found 44% of Ndfa for the Mimosoideae The fixation status of Parapiptadenia zehntneri and Acacia pan- species, a mean that includes high fixing species (Ndfa ¼ 68%) and iculata can not be defined with certainty because, although they species with practically no fixation. Highly fixing species (Ndfa presented non fixing signals, only one was collected from each site greater than 50%, even considering B ¼ –2&) were found in Serra (Serra Talhada and Caruaru, respectively) and there are no registers Talhada, Remı´gio and Caruaru, the most remarkable ones being in literature on their nodulation capacities. Considerations can be Mimosa tenuiflora, Mimosa arenosa and Piptadenia stipulacea.The taken from the taxonomic position of these species. Nodulation has occurrence of these species could be related to the ecological 348 A.D.S. Freitas et al. / Journal of Arid Environments 74 (2010) 344–349 advantage provided by N fixation regarding P uptake in P-limited fixation than the caatinga. In this work, the crucial factor in the low tropical forests (Houlton et al., 2008). They have a much superior fixation estimates was the low number of fixing plants in the performance than Acacia senegal which is cultivated in the semi-arid vegetation (2.4% in Serra Talhada and 11.8% in Remı´gio). An analysis region of Sudan (Raddad et al., 2005), demonstrating the high of many works that quantified density, basal areas or wood volume potential of these species for insertion into agroforestry systems and proportions in the caatinga (Sampaio, 1995) confirm that in the for recuperation of degraded areas. majority of cases, the proportion of fixing legumes is low in mature The lowest value of %Ndfa found in Santa Teresinha could be vegetation similar to the ones included in the present study. Gehr- related to its lower availability of water. However, water availability ing et al. (2005) considered that the BNF in Amazonia was high in alone does not explain the different contributions of BNF in the a secondary forest and low in a mature forest, based on the obser- sites studied, since the highest mean contribution per site was vation that there were a greater number of potentially fixing trees in found in Serra Talhada, which is also located in the physiographical the secondary than in the mature forest. In a similar way, after zone of the Serta˜o, a drier zone than the Agreste. Factors not disturbances, the regeneration of the native caatinga vegetation is evaluated in this work, such as nutrient and micorrhiza availability dominated by fixing species during a great part of the succession (Cardoso and Kuyper, 2006), also affect the BNF and they could be process (Pereira et al., 2003). In latter succession stages, as nitrogen more favorable in Serra Talhada. The importance of BNF for accumulates in the system, fixing species lose their ecological nitrogenous nutrition of the plants also had a direct relationship advantage and non fixing species tend to dominate. In earlier caa- with the N contents. Although the plants of Santa Teresinha are of tinga succession stages, where the biomass is predominantly lower %Ndfa and %N, the plants of Serra Talhada, with high levels of composed of fixing legumes, the total input of N could reach %Ndfa do not differ from them regarding the N contents (Table 4). 130 kg ha1 year1, by adopting the same levels of %Ndfa as found in The N content of the legume leaves differed from the non this work. These quantities are considerable and important in the legume leaves. Among the legumes, the content differed according reestablishment of stocks of N in the areas that are exploited for to the d15N signal (fixing or non fixing legumes). Fixing legumes firewood and slash and burn agriculture. In semi-arid climate had 3.1% of N in their leaves and those from the Remı´gio fragment conditions, pure stands of Acacia senegal, with a 48% capacity of reached 3.9%. Even among non fixing species, the legumes were Ndfa, introduced 36 kg of N ha1 year1at four years of age, in the still richer in N (2.48%) than the non legumes (2.01%). This richness Sudan (Raddad et al., 2005). Although there are no estimates for of N in leaves of legume species, even in non fixing ones, also annual biomass production from Mimosoideae cultivation in the observed in species of the Cerrado and Amazonia (Sprent et al., Brazilian semi-arid, the high nitrogen fixation levels found in this 1996; Martinelli et al., 2000; Bustamante et al., 2004), is attributed work indicate a great potential for considerable inputs of N in such to their growth habit (McKey, 1994), which demands large quan- systems. tities of nitrogen and maximizes the rates of photosynthesis per leaf Very few studies make simultaneous estimates of %Ndfa and area. Our results demonstrate that, in the caatinga, the capacity of primary production of the systems and this is one of the most BNF guarantees even more elevated content of N in the leaves. important problems for estimating total N fixed quantities in The estimated quantities of N added annually by BNF to the leaf communities (Boddey et al., 2000). Thus, quantifications of N added biomass vary from 2.5 kg ha1 year1, in Serra Talhada, to 9.3 kg ha1 to natural ecosystems are scarce in literature (Cleveland et al., year1, in Remı´gio, using the most conservative value of B (-2&). 1999). In spite of the limitations imposed by using estimates of Adopting B ¼ 0&, these values were only slightly higher: 3.3 to biomass fixation taken from data of other works, the quantification 11.2 kg ha1 year1. It has been reported (Ho¨gberg, 1997; Boddey of added N presented in this work has the importance of being the et al., 2000) that the impact of different B values has little signifi- first attempt for caatinga vegetation and one of very few for any cance in the final estimations of BNF in sites where the 15N abun- native vegetation in semi-arid areas and gives an idea of the order dance of reference species is > þ 5& or <5&. In humid tropical of magnitude of annually fixed N through the process of symbiotic forests, the estimates made with different B values may signify fixation in tree legumes. considerable differences in the final fixed N contribution. In a humid forest of French Guyana, Roggy et al. (1999) observed a variation of 5. Conclusion 33% in the estimate of annual N fixation when using B ¼ 0orB¼2&, in a situation where the reference species had a signal of 5.01&. The results suggest that there is a good potential for use of the These estimates indicate that BNF contribution to the leaf 15N natural abundance methodology for evaluating the fixation biomass in well preserved caatinga is small, reaching a little more ability of caatinga tree legumes under field conditions. Not only is 1 1 than 11 kg ha year . The contribution to the whole plant would the identification of N2-fixing species possible but also reasonable be higher considering the N translocated to branches and roots. estimates of %Ndfa can be made. Some legume species of the caa- However, most of the annual accumulation of N is directed to the tinga have the potential of fixing great quantities of nitrogen, in leaves, whose biomass is renovated each year. Therefore, the true particular Mimosa tenuiflora, Mimosa arenosa and Piptadenia stip- contribution would probably be less than 20 kg ha1 year1. ulacea. The mean contributions of BNF for plant nitrogen were high, Although this quantity of N could be considered small, it is the same reaching up to 68%. However, the estimated quantities of N added order of magnitude of estimates for leaf biomasses in two humid annually to the leaf biomass in these systems were low, between forests, carried out by Roggy et al. (1999), in French Guyana 2.5 and 11.2 kg ha1 year1, because of the low proportions of fixing (7 kg ha1 year1). No measurements were found for other semi- plants in the vegetation. In situations of native vegetation regen- arid ecosystems to compare with our estimation. Due to the absence eration, where the succession is dominated by fixing species, N of specific studies, Cleveland et al. (1999) estimated that BNF for fixation could reach 130 kg ha1 year1. Brazilian xeromorphic forests should be of the same order of magnitude as those in deciduous tropical forests (between 7.5 and Acknowledgements 17.6 kg ha1 year1), considering that they have a similar floristic composition. The results for the Remı´gio fragment are on the lower This study was developed with financial support from the limits of these estimates and those for the Serra Talhada fragment Conselho Nacional de Desenvolvimento Cientı´fico e Tecnolo´ gico are even lower. Most tropical and subtropical deciduous forests (CNPq Universal Processo 479345/2006-6 and IMSEAR Edital MCT/ present more favorable moisture and soil fertility conditions for N CNPq 01/2005) and the Inter American Institute for Global Change A.D.S. Freitas et al. / Journal of Arid Environments 74 (2010) 344–349 349

Research (IAI – CRN 001). The authors are grateful to the Empresa Gehring, C., Vlek, P.L.G., Souza, L.A.G., Denich, M., 2005. Biological nitrogen fixation Pernambucana de Pesquisa Agropecua´ria (IPA) and Universidade in secondary regrowth and mature rainforest of central Amazonia. Agriculture Ecosystems and Environment 111, 237–252. Federal de Campina Grande for access of vegetation areas at the Handley, L.L., Odee, D., Scrimgeour, C.M., 1994. d15N and d13C patterns in savanna experimental stations of Caruaru, Serra Talhada and Santa vegetation: dependence on water availability and disturbance. Functional Teresinha. Ecology 8, 306–314. Ho¨gberg, P., 1997. 15N natural abundance in soil-plant systems. New Phytologist 137, 179–203. References Houlton, B.Z., Wang, Y.P., Vitousek, P.M., Field, C.B., 2008. A unifying framework for dinitrogen fixation in the terrestrial biosphere. Nature 454, 327–331. Alcoforado-Filho, F.G., Sampaio, E.V.S.B., Rodal, M.J.N., 2003. Florı´stica e fitossocio- Kauffman, J.B., Sanford, R.L., Cummings, D.L., Salcedo, I.H., Sampaio, E.V.S.B., 1993. logia de um remanescente de vegetaça˜o caducifo´ lia espinhosa arbo´ rea em Biomass and nutrient dynamics associated with slash fires in neotropical dry Caruraru, Pernambuco. Acta botanica brasilica 17, 287–303. forests. Ecology 74, 140–151. Allen, O.N., Allen, E.K., 1981. The Leguminosae: A Source Book of Characteristics Use Martinelli, L.A., Almeida, S., Brown, I.F., Moreira, M.Z., Victoria, R.L., Filoso, C.S., and Nodulation. University of Wisconsin Press, Wisconsin, pp. 812. Ferreira, A.C., Thomas, W.W., 2000. Variation in nutrient distribution and Boddey, R.M., Peoples, M.B., Palmer, B., Dart, P., 2000. Use of the 15N natural potential nutrient losses by selective logging in a Humid Tropical Forest of abundance technique to quantify biological nitrogen fixation by woody Rondoˆnia. Brazil. Biotropica 32, 597–613. perennials. Nutrient Cycling in Agroecosystems 57, 235–270. McKey, D., 1994. Legumes and nitrogen: the evolutionary ecology of a nitrogen- Bustamante, M.M.C., Martinelli, L.A., Silva, L.A., Camargo, P.B., Klink, C.A., demanding lifestyle. In: Sprent, J.I., McKey, D. (Eds.), Advances in Legume Sis- Domingues, T.F., Santos, R.V., 2004. 15N natural abundance in woody plants and tematics 5: The Nitrogen Factor. Royal Botanic Gardens, Kew, pp. 211–228. soils of central Brazilian savannas (Cerrado). Ecological Applications 14, Ndiaye, M., Ganry, F., 1997. Variation in the biological N2 fixation by tree legumes in 200–213. three ecological zones from the North to the south of Senegal. Arid Soil Cardoso, I.M., Kuyper, T.W., 2006. Mycorrhizas and tropical soil fertility. Agriculture, Research and Rehabilitations 11, 245–254. Ecosystems and Environment 116, 72–84. Pereira, I.M., Andrade, L.A., Barbosa, M.R.V., Sampaio, E.V.S.B., 2002. Composiça˜o Cleveland, C.C., Townsend, A.R., Schimel, D.S., Fisher, H., Howarth, R.W., Hedin, L.O., florı´stica e ana´lise fitossociolo´ gica do componente arbustivo-arbo´ reo de um Perakis, S.S., Latty, E.F., Von Fischer, J.C., Elseroad, A., Wasson, N.F., 1999. Global remanescente florestal no Agreste paraibano. Acta Botanica Brası´lica 16, 357–369. patterns of terrestrial biological nitrogen (N2) fixation in natural ecosystems. Global Biogeochemical Cycles 13, 623–645. Pereira, I.M., Andrade, L.A., Sampaio, E.V.S.B., Barbosa, M.R.V., 2003. Use-history Dantas, J.S., 2003. Dinaˆmica da produça˜o e decomposiça˜o de folhedo e ciclagem de effects on structure and flora of Caatinga. Biotropica 35, 154–165. nutrientes em um ecossistema de caatinga arbo´ rea no Agreste da Paraı´ba. Queiroz, L.P., 2006. The Brazilian Caatinga: Phytogeografical patterns inferred from Graduate monograph, Universidade Federal da Paraı´ba Areia, Paraı´ba, Brasil 37p. distribution data of the Leguminosae. In: Pennington, R.T., Lewis, G.P., Faria, S.M., Lima, H.C., 1998. Additional studies of the nodulation status of legume Ratter, J.A. (Eds.), Neotropical Savannas and Seasonally Dry Forests. Plant species in Brazil. Plant and Soil 200, 185–192. Diversity, Biogeography and Conservation. CRC Press, New York, pp. 121–157. Faria, S.M., Franco, A.A., Jesus, R.M., Menandro, M.S., Baitello, J.B., Mucci, E.S.F., Raddad, A.Y., Salih, A.A., El Fadl, M., Kaarakka, V., Luukkanen, O., 2005. Symbiotic Dobereiner, J., Sprent, J.I., 1984. New nodulating legume trees from southeast nitrogen fixation in eight Acacia senegal provenances in dryland clays of the 15 Brazil. New Phytologist 98, 317–328. Blue Nile Sudan estimated by the N natural abundance method. Plant and Soil Faria, S.M., Lima, H.C., Franco, J.B., Mucci, E.S.F., Sprent, J.I., 1987. Nodulation of 275, 261–269. legume trees from southeast Brazil. Plant and Soil 99, 347–356. Roggy, J.C., Pre´vost, M.F., Gourbiere, F., Casabianca, H., Garbaye, J., Domenach, A.M., 15 Faria, S.M., Lima, H.C., Carvalho, A.M., Gonçalves, V.F., Sprent, J.I., 1994. Occurrence 1999. Leaf natural N abundance and total N concentration as potential indi- of nodulation in legume species from Bahia, Minas Gerais and Espı´rito Santo cators of plant nutrition in legumes and pioneer species in a rain forest of states of Brazil. In: Sprent, J.I., Mcmey, D. (Eds.), Advances in Legume System- French Guiana. Oecologia 120, 171–182. atics 5. The nitrogen factor. Royal Botanic Gardens, Kew, pp. 17–23. Sampaio, E.V.S.B., 1995. Overview of the Brazilian caatinga. In: Bullock, S.H., Ferraz, E.M.N., Rodal, M.J., Sampaio, E.V.S.B., 2003. Physiognomy and structure of Mooney, H.A., Medina, E. (Eds.), Seasonally Dry Tropical Forests. Cambridge vegetation along an altitudinal gradient in the semi-arid region of northeastern University Press, Cambridge, pp. 35–63. Brazil. Phytocoenologia 33, 71–92. Schulze, E.D., Gebauer, G., Ziegler, H., Lange, O.L., 1991. Estimates of nitrogen fixation Freitas, A.D.S., 2008. Abundaˆncia natural do 15N e fixaça˜o biolo´ gica do N2 em by trees on an aridity gradient. Oecologia 88, 451–455. espe´cies arbo´ reas da caatinga. Doctors Dissertation, Universidade Federal de Shearer, G., Kohl, D.H., 1986. N2-fixation in field settings: estimations based on 15 Pernambuco, Recife, Brasil, 77p. natural N abundance. Australian Journal of Plant Physiology 13, 699–756. Gathumbi, S.M., Cadisch, G., Giller, K.E., 2002. 15N natural abundance as a tool for Shearer, G., Kohl, D.H., Virginia, R.A., Bryan, B.A., Skeeters, J.L., Nilsen, E.T., assessing N -fixation of herbaceous, shrub and tree legumes in improved Sharife, M.R., Rundel, P.W., 1983. Estimates of N2-fixation from variation in the 2 15 fallows. Soil Biology and Biochemistry 34, 1059–1071. natural abundance of N in Sonoran Desert ecosystems. Oecologia 56, 365–373. 15 Gehring, C., Vlek, P.L.G., 2004. Limitations of the 15N natural abundance method for Sprent, J.I., Geoghegan, I.E., Whitty, P.W., James, E.K., 1996. Natural abundance of N 13 estimating biological nitrogen fixation in Amazonian forest legumes. Basic and and C in nodulated legumes and other plants in the cerrado and neighbouring Applied Ecology 5, 567–580. regions of Brazil. Oecologia 105, 440–446.