INTRODUCTION
Water and nutrients are basic resources for growth and activity of soil microbial communities (Atlas & Bartha Reference ATLAS and BARTHA2002). The availability of C and N favours heterotrophic micro-organisms specialized on labile nutrients, whereas low availability in these resources favours microbes that are able to break down more recalcitrant soil organic matter (Fontaine & Barot Reference FONTAINE and BAROT2005, Fontaine et al. Reference FONTAINE, MARIOTTI and ABBADIE2003, Hu et al. Reference HU, VAN BRUGGEN and GRÜNWALD1999). Consequently, microbial succession during soil organic matter decomposition depends on the enzymatic capabilities or resource specificity of the different micro-organisms involved in this process (Balser et al. Reference BALSER, KINZIG, FIRESTONE, Kinzing, Pacala and Tilman2002, Paul & Clark Reference PAUL and CLARK1989). Other micro-organisms such as ammonia-oxidizing bacteria do not depend on organic matter directly, but differences in the availability of organic resources for heterotrophs may mediate N mineralization that provides them with ammonium (Booth et al. Reference BOOTH, STARK and RASTETTER2005, Verhagen & Laanbroek Reference VERHAGEN and LAANBROEK1991). Altogether this suggests that both type and supply of resources may determine the structure of soil microbial communities.
The diversity and functioning of soil microbial communities is still poorly understood, despite the critical role of soil microbes in the regulation of nutrient cycling and ecosystem functioning (Balser et al. Reference BALSER, KINZIG, FIRESTONE, Kinzing, Pacala and Tilman2002). Some studies indicate that soil micro-organisms are ubiquitous (Finlay Reference FINLAY2002, Finlay & Clark Reference FINLAY and CLARK1999), which may suggest that resource availability has little effect on microbial diversity and function, while other studies show that the size and composition of the microbial communities vary according to resource availability (Balser & Firestone Reference BALSER and FIRESTONE2005, Carney & Matson Reference CARNEY and MATSON2005, Carney et al. Reference CARNEY, MATSON and BOHANNAN2004, Cookson et al. Reference COOKSON, ABAYE, MARSCHNER, MUPHY, STOCKDALE and GOULDING2005, Reference COOKSON, OSMAN, MARSCHNER, ABAYE, CLARK, MUPHY, STOCKDALE and WATSON2007; Waldrop et al. Reference WALDROP, BALSER and FIRESTONE2000). Microbial communities are therefore likely dependent on the environmental context and major limiting factors of each ecosystem.
Rainfall seasonality and habitat heterogeneity alter the availability of resources (water and nutrients) in tropical deciduous forests by affecting soil moisture and organic matter input (García-Oliva et al. Reference GARCÍA-OLIVA, SVESHTAROVA and OLIVA2003, Roy & Singh Reference ROY and SINGH1994). It has been suggested that water limits C and N mineralization and nitrification in these forests (García-Méndez et al. Reference GARCÍA-MÉNDEZ, MAASS, MATSON and VITOUSEK1991, García-Oliva et al. Reference GARCÍA-OLIVA, SVESHTAROVA and OLIVA2003, Singh & Kashyap Reference SINGH and KASHYAP2006, Singh et al. Reference SINGH, RAGHUBANSHI, SINGH and SRIVASTAVA1989), and recently that C availability is critical for C and N transformations (Montaño et al. Reference MONTAÑO, GARCÍA-OILVA and JARAMILLO2007). Despite the fact that tropical deciduous forest covers around 42% of the world's tropical areas and around 47% of the forested area in Latin America (Murphy & Lugo Reference MURPHY and LUGO1986) there are few studies on bacterial ecology (Jha et al. Reference JHA, SINGH and KASHYAP1996, Noguez et al. Reference NOGUEZ, ARITA, ESCALANTE, FORNEY, GARCÍA-OLIVA and SOUZA2005, Reference NOGUEZ, ESCALANTE, FORNEY, NAVA-MENDOZA, ROSAS, SOUZA and GARCÍA-OLIVA2008; Singh & Kashyap Reference SINGH and KASHYAP2006) in this highly threatened ecosystem (Miles et al. Reference MILES, NEWTON, DEFRIES, BLYTH, KAPOS and GORDON2006). Noguez et al. Reference NOGUEZ, ARITA, ESCALANTE, FORNEY, GARCÍA-OLIVA and SOUZA(2005) reported, using TRFLP, that neighbouring hilltops and hillslopes within the tropical deciduous forest at the Chamela-Cuixmala Biosphere Reserve in Jalisco, Mexico shared only 54% of their bacterial Operational Taxonomic Units (OTU). Additionally, Noguez et al. (Reference NOGUEZ, ESCALANTE, FORNEY, NAVA-MENDOZA, ROSAS, SOUZA and GARCÍA-OLIVA2008) found correlations between numbers of OTUs and soil N transformation in this ecosystem associated with two soil-aggregate size fractions. This suggested that the richness and composition of bacterial communities could be related to the availability of nutrients, which has been found to be different in these topographic positions (Montaño et al. Reference MONTAÑO, GARCÍA-OILVA and JARAMILLO2007). However, there is little information in tropical deciduous forests about the links between resource availability and how variation in microbial communities may regulate soil biogeochemical processes in tropical forests. Despite it being well established that culturable bacteria represent only a small percentage (1–10%) of the bacteria species inhabiting soil (Rothschild Reference ROTHSCHILD2006, Torsvik & Øvreås Reference TORSVIK and ØVREÂS2002), bacterial cultivation techniques in combination with biochemical (fatty acids) measurements could be an inexpensive, complementary approach to TRFLP in order to identify bacterial species and make comparative studies of species distribution (Nichols Reference NICHOLS2007).
This research was conducted to study some potential links among microbial activity, culturable bacterial communities and resource availability in two topographic positions with known differences in nutrient availability (hilltops and hillslopes) in a tropical deciduous forest in western Mexico. This approach provided an experimental setup to assess the response of microbial activity to natural differences in soil resources without soil disturbance effects or artificial manipulations. Our main hypothesis was that soil microbial biomass, as well as culturable heterotrophic and nitrifying bacteria would be higher in soils with high availability of soil resources and that the variation in bacterial communities would correlate with C mineralization and nitrification processes.
METHODS
Study area
The study area is located at the Chamela-Cuixmala Biosphere Reserve, near the Pacific Coast of Mexico (19° 29′ N, 105° 01′ W). Mean annual temperature is 24.6 °C and mean annual rainfall is 746 mm (1983–2004). Rainfall is mainly concentrated in a clearly marked wet season that lasts from June to October, and peaks in September (García-Oliva et al. Reference GARCÍA-OLIVA, CAMU, MAASS, Noguera, Vega, García and Quezada2002). The soils are poorly developed sandy clay loams, classified as Eutric Regosols in the FAO system (Cotler et al. Reference COTLER, DURÁN, SIEBE, Noguera, Vega, García and Quezada2002). The soil parent material is Tertiary rhyolite and the dominant clay is kaolinite (Campo et al. Reference CAMPO, MAASS and DE PABLO2001). Soil organic matter (SOM) content is < 5% and is mainly concentrated in the top 5 cm (García-Oliva & Maass Reference GARCÍA-OLIVA and MAASS1998). The vegetation is a highly diverse tropical deciduous forest (Lott Reference LOTT1993), where most tree species are leafless during the 7-mo dry season (Martínez-Yrízar et al. Reference MARTÍNEZ-YRÍZAR, MAASS, PÉREZ-JIMÉNEZ and SARUKHÁN1996).
Soil sampling
Soil samples were collected from an undisturbed forest located in three contiguous small watersheds that have been extensively studied for a long-term project on ecosystem function (Maass et al. Reference MAASS, JARAMILLO, MARTÍNEZ-YRÍZAR, GARCÍA-OLIVA, PÉREZ-JIMÉNEZ, SARUKHÁN, Noguera, Vega, García and Quezada2002). These watersheds have the same geological age, parent material and different topographic units (slope and aspect; López-Blanco et al. Reference LÓPEZ-BLANCO, GALICIA and GARCÍA-OLIVA1999). We collected soil from two topographic positions (sites): hilltops (slope = 1.2° ± 0.7°) and south-facing hillslopes (slope = 26° ± 3°), distributed in the three watersheds. Soils have different concentrations of organic C (hilltop: 37 mg C g−1, hillslope: 24 mg C g−1) and dissolved organic C (244 μg C g−1 and 92 μg C g−1 for hilltop and hillslope, respectively; Montaño et al. Reference MONTAÑO, GARCÍA-OILVA and JARAMILLO2007). Both sites have similar vegetation (Balvanera et al. Reference BALVANERA, LOTT, SEGURA, SIEBE and ISLAS2002), annual solar radiation index (4356 MJ m−2 y−1 and 4426 MJ m−2 y−1 for hilltop and hillslope, respectively; Galicia et al. Reference GALICIA, LÓPEZ-BLANCO, ZARCO-ARISTA, FILIPS and GARCÍA-OLIVA1999), soil (Eutric Regosol) with similar texture, and water-holding capacity (Galicia et al. Reference GALICIA, LÓPEZ-BLANCO, ZARCO-ARISTA, FILIPS and GARCÍA-OLIVA1999).
Ten replicate sampling plots (10 × 15 m) were established in each of the two topographic positions (sites). The plots were located at least 300 m apart. Soil samples were collected in each plot on three sampling dates, in the dry (April), early rainy (June) and rainy (September) seasons in 2004, which had an annual precipitation of 578 mm. Fifteen undisturbed topsoil samples (0–5 cm depth), where most root and microbial activity is concentrated, were randomly collected from each plot at each sampling date. The samples were thoroughly mixed to form a composite soil sample per plot. Each soil sample was passed through a 2-mm sieve and one subsample was oven dried at 75 °C to determine soil moisture gravimetrically. The remaining soil was refrigerated at 10 °C until processing to prevent microbial proliferation and was processed in the laboratory within 10 d of sampling.
Soil chemical analyses
Soil pH was measured in deionized water (w:v 1:2) with an electric digital pH meter (Corning). Dried soil samples were ground with a pestle and mortar prior to total soil nutrient analyses. Total C was determined by combustion and coulometric detection. Dissolved organic carbon (DOC) was extracted in deionized water, filtered through a Whatman No. 42 paper (Jones & Willett Reference JONES and WILLETT2006), and determined with a C analyser. Total N and P were determined after acid digestion by a macro-Kjeldahl method. Total N was determined by colorimetrical analysis and total P by the molybdate colorimetric method after ascorbic acid reduction (Murphy & Riley Reference MURPHY and RILEY1962). Inorganic N (NH4+ and NO3−) was extracted with 2 M KCl, followed by filtration through a Whatman No. 1 paper filter (Robertson et al. Reference ROBERTSON, COLEMAN, BLEDSOE and SOLLINS1999), and determined colorimetrically by the phenol-hypochlorite method. Inorganic phosphorus was extracted with sodium bicarbonate (Pi; pH 8.5), and was determined by the molybdate-ascorbic acid method (Murphy & Riley Reference MURPHY and RILEY1962). All C forms were determined with a Carbon Analyser UIC Mod. CM5012, while N and P forms were determined colorimetrically using a Bran-Luebbe Auto Analyser III (Norderstedt, Germany).
Soil functional processes
Two microbial functional processes were considered: potential C mineralization and net nitrification. These microbial processes were measured in 16-d laboratory aerobic incubations. Samples of homogenized fresh soil were wetted to field capacity, placed in polyvinyl-chloride (PVC) tube cores and incubated in jars at 26 °C. The jars were regularly aerated for 1-h periods and every 48 h they were watered by weight to field capacity with deionized water (Robertson et al. Reference ROBERTSON, COLEMAN, BLEDSOE and SOLLINS1999). Potential C mineralization was estimated as evolved CO2-C collected in 1 M NaOH traps, which were changed every 2 d. Carbonates were precipitated by adding 1.5 M BaCl2 and then titrated with 1 M HCl. Net nitrification during each incubation period was estimated as the difference between initial and final nitrate concentrations, which were analysed using the method described above.
Soil microbiological analyses
Microbial biomass C was determined with the chloroform fumigation-extraction method (Vance et al. Reference VANCE, BROOKES and JENKINSON1987). Fumigated and non-fumigated samples were incubated for 24 h at 25 °C and constant moisture (Brookes et al. Reference BROOKES, LANDMAN, PRUDEN and JENKINSON1985). Microbial C was extracted from both fumigated and non-fumigated samples with 0.5 M K2SO4, filtered through a Whatman No. 42 paper, and measured using a C analyser (UIC, mod. CM5012). Microbial C was calculated by subtracting the extracted C in non-fumigated samples from that of fumigated samples and dividing by a KEC value of 0.45 (Joergensen Reference JOERGENSEN1996).
Culturable heterotrophic and nitrifying bacteria were quantified using 10 g of fresh soil from each composite sample and processing under sterile conditions. The soil was shaken with 90 ml of sterile deionized water, and aliquots (0.1 ml) of three ten-fold dilution series were used as inoculum to determine the culturable heterotrophic bacteria with the plate count method (Zuberer Reference ZUBERER and Weaver1994), and the nitrifying bacteria with the most probable number method (MPN; Alexander Reference ALEXANDER and Black1982). For the quantification of culturable heterotrophic bacteria, 0.1 ml of the extract was transferred to plates with sterile Tryptic-Soy-Agar medium (TSA; Bioxon). Plates were incubated at 25 °C and colony-forming units (CFU) were counted after 3 d. Size, shape, colour, concavity, consistency and edge type of colonies of heterotrophic bacteria were used to characterize and group isolates morphologically, which allowed us to identify morphotypes in each plate. Colonies of each morphotype found were streaked repeatedly on TSA medium and incubated again at 25 °C for 80 h to obtain pure cultures. Pure cultures were then transferred into tubes containing solid TSA medium and stored at −40 °C until further biochemical identification based on fatty acid profiles. The ammonium oxidizer bacteria were quantified and determined using ammonium-calcium carbonate ((NH4)2SO4, 0.5 g; K2PO4, 1.0 g; FeSO4 · 7H2O, 0.03 g; NaCl, 0.3 g; MgSO4 · 7H2O, 0.3 g; CaCO3, 7.5 g; deionized sterile water) in agar culture medium (Bioxon). Five replicate culture tubes were used for each dilution. Each tube was inoculated with an extract of 0.1 ml and incubated for 6 wk at 30 °C. After the incubation period, each tube was tested for nitrite using Griess–Ilosvay reagent. The number of positive and negative tubes was used to calculate the CFU using a MPN table (Alexander Reference ALEXANDER and Black1982). We were not able to isolate nitrifying bacteria for identification by fatty acid methyl-esters (FAME).
Whole-cell FAMEs were used for biochemical identification of heterotrophic bacterial isolates. Gram positive and negative bacteria were estimated from the content of branched-chain or monoenoic and cyclopropane fatty acids, respectively. A loop full of cell material of late-log-phase cells of each isolate was transferred to a glass tube and fatty acids were extracted using standard and recommended procedures for gas chromatographic FAME analysis as described by Sasser (Reference SASSER, Klement, Rudolph and Sands1990). Analysis was performed with an Agilent 6890 Plus Chromatograph and the Sherlock system software using the method TSBA41 and the library TSBA41 (Microbial Identification System MIDI Inc, Delaware, USA). A similarity index threshold of 0.30 was applied for acceptance of identification as recommended by the MIDI manual. Some isolates were not found in the identification library, but were considered as different bacterial morphotypes since they had a different composition of fatty acids. All bacterial isolates were incorporated to the bacteria strain collection of the Research Centre Flakkebjerg in Denmark.
Culturable bacterial-species richness was estimated in terms of the observed species and morphotypes number (Sobs; Magurran Reference MAGURRAN2004), which had different composition of fatty acids but rendered no species name using MIDI. We used a dissimilarity index by cluster analysis of the species composition between sites (hilltop and hillslope) and sampling dates (seasons) to estimate the proportion of bacterial species shared between sites. For this analysis we used Ward's Method with Euclidean distances based on species presence-absence data. The index is equal to 0 in cases of complete similarity, that is when two sets of species are identical, and 100% if the sites have no species in common (Magurran Reference MAGURRAN2004).
Statistical analyses
All data were expressed on a dry-weight basis unless otherwise stated. Data were subjected to a repeated-measures analysis of variance (RMANOVA), where site (hilltop or hillslope) was considered as the main effect and sampling dates were treated as repeated measures. A Greenhouse–Geisser correction was estimated for the sphericity assumption of the time factor (von Ende Reference VON ENDE, Scheiner and Gurevitch1993), but it did not modify the significance of any P values. A Tukey's HSD multiple comparison test was used when statistical differences were observed with RMANOVA (Sokal & Rohlf Reference SOKAL and ROHLF1995). The data were transformed (logarithm or square root) to satisfy ANOVA assumptions when required (Sokal & Rohlf Reference SOKAL and ROHLF1995), however, results are reported in their original scale of measurement. A stepwise regression analysis was used to assess the importance of available nutrients in relation to microbial biomass and bacterial communities and of both nutrients and bacterial communities on soil C mineralization and nitrification. All statistical analyses were performed with Statistica ver. 6.0 software (StatSoft, Tulsa, USA), and a P ≤ 0.05 was considered significant.
RESULTS
Soils had a slightly acid pH and hillslope soils had a lower pH than the hilltop soils in the dry and the early rainy seasons (Table 1). Hilltop soils had higher soil moisture than the hillslope soils. In both sites, the maximum soil moisture was recorded in the early rainy season (Table 1). Hilltop soils had higher total C than the hillslope soils. The total N and P concentrations and C:P ratios were not affected by site or sampling date, but the hilltops had higher C:N ratios than the hillslopes (Table 1).
Table 1. Mean ± SE for pH, moisture, total nutrients, and C:nutrient ratios for hilltops and hillslopes in a tropical deciduous forest of Chamela Jalisco, Mexico. Values followed by a different uppercase letter (A and B) indicate that means are significantly different (P ≤ 0.05) between sites (hilltop vs hillslope) within sampling date. Similarly, different lowercase letters (a, b, c) indicate that means are significantly different (P ≤ 0.05) among sampling dates within a site. RMANOVA F-ratios with: ns = not significant. * P < 0.05; ** P < 0.01 and *** P < 0.001.
Dissolved organic carbon (DOC) concentrations were in general higher at the hilltop soils than at the hillslope soils (Table 2). A significant date × site interaction indicated that DOC was more variable among sampling dates in the hilltop than in the hillslope soils (Table 2) though it showed the same trend decreasing from the dry to the early rainy season and increasing to the middle of the rainy season. Ammonium concentrations were also higher in the hilltop than in the hillslope soils on the three sampling dates, and decreased progressively from the dry towards the rainy season in both sites. In contrast, nitrate concentrations did not vary between sites and were lowest in the dry season. The hillslope soils had higher NO3:NH4 ratios than the hilltop soils, with the highest NO3:NH4 ratios in the rainy season (Table 2). There was a significant interaction for extractable phosphorus concentration indicating that the Pi availability changed in a different way with time for both soils, with no clear trend suggesting more P availability in a given site or sampling date (Table 2). Hilltop soils had higher potential C mineralization (CO2-C evolved) than the hillslope soils, and the samples collected in the dry season had the highest potential C mineralization (Figure 1a). Net nitrification was consistently greater for the hillslope soils than for the hilltop soils, and the highest nitrification was recorded in samples collected in the early rainy and rainy seasons (Figure 1b).
Table 2. Mean ± SE for seasonal variation in dissolved organic carbon, ammonium, nitrate, and bicarbonate extractable phosphorous concentrations, and nitrate:ammonium ratios for hilltops and hillslopes in a tropical deciduous forest of Chamela Jalisco, Mexico. Values followed by a different uppercase letter (A and B) indicate that means are significantly different (P ≤ 0.05) between sites (hilltop vs hillslope) within sampling date. Similarly, different lowercase letters (a, b, c) indicate that means are significantly different (P ≤ 0.05) among sampling dates within a site. DOC = dissolved organic carbon, Pi = inorganic phosphorus extracted with sodium bicarbonate. RMANOVA F-ratios with: ns = not significant. * P < 0.05; ** P <0.01 and *** P < 0.001.
Figure 1. Seasonal variation (mean ± SE) of potential C mineralization rate (CO2-C) (a), and net nitrification (b) after a 16-d incubation period in hilltop and hillslope soils (high C and low C respectively) from a tropical deciduous forest at Chamela, Mexico. Different uppercase letters (A and B) indicate that means are significantly different (P ≤ 0.05) between sites (hilltop vs hillslope) within sampling date (season). Similarly, different lowercase letters (a, b, c) indicate significant differences (P ≤ 0.05) among sampling dates within a site. RMANOVA F-ratios with: ns = not significant, * P ≤ 0.05; ** P ≤ 0.01, *** P ≤ 0.001.
Microbial C was higher in the hilltop soils than at the hillslope soils, and decreased from the dry season to the rainy season in both sites (Figure 2a). The colony-forming units (CFU) of culturable heterotrophic bacteria were higher in the hilltop than in the hillslope soils in the early rainy season. In both sites the CFU of heterotrophic bacteria were lowest in the rainy season, and only in the hilltop soils a peak was observed in the early rainy season (Figure 2b). In contrast, the most probable number (MPN) of nitrifying bacteria was higher in the hillslope than in the hilltop soils only in the rainy season, and increased from the dry season to the rainy season in both sites (Figure 2c).
Figure 2. Seasonal variation (mean ± SE) in microbial carbon and culturable bacterium counts (CFU) (a), heterotrophs (b) and nitrifiers (c) in hilltop and hillslope soils (high C and low C respectively) in a tropical deciduous forest of Chamela, Mexico. Different uppercase letter (A and B) indicate that means are significantly different (P ≤ 0.05) between sites (hilltop vs hillslope) within sampling date (season). Similarly, different lowercase letters (a, b, c) indicate significant differences (P ≤ 0.05) among sampling dates within a site. RMANOVA F-ratios with: ns = not significant, * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001.
Culturable bacterium species richness was higher at the hilltop than at the hillslope soils in the dry and early rainy seasons, but the sites had similar bacterial richness in the middle of the rainy season (Site × Date, F = 5.3, P < 0.01, Appendix 1). The highest culturable bacterium richness was found in the rainy season in both sites. We documented a total of 28 genera with 61 bacterial species in the two sites. Nineteen species were common to both sites and 47% of the total species was found only in the hilltop and 31% only in the hillslope soils. Bacillus, Brevibacillus and Paenibacillus species accounted for 31% of the bacterial isolates found. From the total number of culturable heterotrophic bacterial species, 67% were Gram-positive and 33% were Gram-negative bacteria (Appendix 1). The lowest dissimilarity index values (23%) of the culturable bacterium composition were found at the hillslopes of the dry and early rainy seasons, while the hilltops of the early rainy and rainy seasons had the highest dissimilarity (58%; Figure 3).
Figure 3. Dissimilarity cluster (Ward's method, based on Euclidean distances) showing differences in heterotrophic bacterial composition in the hilltop and hillslope soils (high C and low C respectively) collected in Dry, Early rainy (E-rainy) and Rainy seasons (sampling date) in a tropical deciduous forest of Chamela, Mexico.
Stepwise regression analysis to assess the importance of available nutrients in relation to microbial biomass and bacterial communities and of both nutrients and bacterial communities on soil C mineralization and nitrification showed some significant but in general weak correlations among most variables (Table 3). In general, both DOC and ammonium were positively related to microbial biomass (C microbial), culturable heterotrophic bacteria, and potential C mineralization, but inversely related to net nitrification and the culturable nitrifying bacteria.
Table 3. Stepwise multiple regressions, correlation coefficients and significance levels to detect controls on microbial biomass, culturable bacterium communities and microbial activity in a tropical deciduous forest of Mexico. Regression analyses were performed on pooled data across the sites and seasons (forward procedure with n = 60). In the analyses all soil variables were incorporated in the model to detect the main controls on Y-variable with two exceptions: for microbial variables both mineralization and nitrification were excluded and for both C mineralization and nitrification the manipulated variables during the incubation were not considered. aDOC = dissolved organic carbon, HB = culturable heterotrophic bacteria. Significant: * P < 0.05; ** P < 0.01 and *** P < 0.001.
DISCUSSION
The results suggested that resource availability is an important controlling factor for both microbial growth and culturable bacterial species richness in this tropical deciduous forest. Dissolved organic carbon and ammonium accumulation on the hilltops compared with hillslopes correlated with higher microbial growth most likely due to higher availability of organic C for heterotrophic micro-organisms as previously suggested (García-Oliva et al. Reference GARCÍA-OLIVA, SVESHTAROVA and OLIVA2003, Montaño et al. Reference MONTAÑO, GARCÍA-OILVA and JARAMILLO2007). The increase in nitrifying bacterial colonies in soil with less ammonium and DOC in the rainy season suggests that low C availability reduced both growth and dominance of the heterotrophic bacteria and favoured nitrifiers. Nitrifiers obtain energy from ammonium oxidation, not from organic C (Booth et al. Reference BOOTH, STARK and RASTETTER2005, Verhagen & Laanbroek Reference VERHAGEN and LAANBROEK1991), and this could explain the decline in ammonium and the increase in the NO3:NH4 ratio and net nitrification in the hillslope soils, as suggested in a previous study conducted in this forest (Montaño et al. Reference MONTAÑO, GARCÍA-OILVA and JARAMILLO2007). However, there were no significant differences in nitrate concentrations, and this might be explained by leaching or denitrification increasing variability. Additionally, it is also possible that our results did not reflect total nitrate production due to the limitation of net measurements. Overall, it seems that, as hypothesized, the differences in C mineralization and nitrification between our study sites could be due to the presence of different bacteria, which in turn could be regulated by C availability (Cleveland et al. Reference CLEVELAND, NEMERGUT, SCHMIDT and TOWNSEND2007, Cookson et al. Reference COOKSON, OSMAN, MARSCHNER, ABAYE, CLARK, MUPHY, STOCKDALE and WATSON2007, Fierer et al. Reference FIERER, BRADFORD and JACKSON2007).
Similarly, the higher richness of culturable heterotrophic bacterial species on the hilltops could be also associated to the abundance of resources, because it may provide a broader range of different niches to be occupied by more bacterial species (Begon et al. Reference BEGON, HARPER and TOWNSEND1986, Diamond Reference DIAMOND1988, Torsvik et al. Reference TORSVIK, ØVREÂS and THINGSTAD2002). This agrees with studies in other contexts showing that soils with more resource availability have higher microbial diversities (Bossio et al. Reference BOSSIO, SCOW, GUANAPALA and GRAHAM1998, Torsvik & Øvreås Reference TORSVIK and ØVREÂS2002, Zhou et al. Reference ZHOU, XIA, TRAVES, WUL, MARCH, O'NEILL, PALUMBO and TIEDJE2002). Nevertheless, further work is required to test this hypothesis because our measurements cannot resolve if DOC and nutrients were a substrate or a product of bacterial richness and microbial activity, in order to establish causation.
Bacterial richness showed an opposite seasonal trend to microbial biomass that may be explained in different ways: (1) a few active and fast-growing bacteria are able to use available nutrients when water availability is low, and (2) bacterial richness increases in the rainy season but growth is limited by low DOC and nutrient availability, and reduced by grazing. The increase in microbial C in the dry season, when an accumulation of DOC, P and NH4+ also occurs, has already been documented in tropical deciduous forests (Campo et al. Reference CAMPO, JARAMILLO and MAASS1998, Montaño et al. Reference MONTAÑO, GARCÍA-OILVA and JARAMILLO2007, Singh et al. Reference SINGH, RAGHUBANSHI, SINGH and SRIVASTAVA1989). This may be due to the presence of fast-growing bacteria that are resistant to drought conditions and have high efficiency in the utilization of labile resources (Atlas & Bartha Reference ATLAS and BARTHA2002). The rainy-season samples with lower labile resource availability, in contrast, showed higher species richness but lower growth. It is likely that these bacteria use most of their resources for complex organic matter decomposition to obtain nutrients at the expense of growth and reproduction (Atlas & Bartha Reference ATLAS and BARTHA2002, Fontaine et al. Reference FONTAINE, MARIOTTI and ABBADIE2003, Waldrop et al. Reference WALDROP, BALSER and FIRESTONE2000). The main functional implication of the impairment of microbial biomass and bacterial richness is that the conservation of nutrients in microbial biomass during the dry season (Singh et al. Reference SINGH, RAGHUBANSHI, SINGH and SRIVASTAVA1989) would depend on a few bacterial species functioning well under high nutrient, but low water availability.
The predominance of Gram-positive bacteria in our study sites could be explained by the seasonal fluctuations in water and nutrients that occur in this forest, which may act as stress factors that select spore-forming Gram-positive bacteria, more resistant to high stress and heterogeneous, unpredictable conditions (Atlas & Bartha Reference ATLAS and BARTHA2002, Paul & Clark Reference PAUL and CLARK1989, Williams & Rice Reference WILLIAMS and RICE2007). Gram-positive bacteria are considered as more capable of metabolizing complex C substrates than Gram-negative bacteria, which are known for being fast-growing and highly competitive for simple substrates (Paul & Clark Reference PAUL and CLARK1989, Waldrop et al. Reference WALDROP, BALSER and FIRESTONE2000). The higher abundance of Gram-positive bacteria in these soils is supported by data obtained using fatty acid biomarkers, a culture-independent quantitative measure of soil microbial groups, in a microbial community study conducted later in the same plots (Gavito et al. unpubl. data).
The low similarity between sites in bacterial species composition may indicate that not all heterotrophic bacteria respond in the same way to resource availability. Studies conducted in the same forest, revealed also that hilltops had higher bacterial diversity than hillslopes, and had 36% and 26%, respectively, of unique bacterial Operational Taxonomic Units (OTU) using terminal restriction fragment-length polymorphism (TRFLPs) (Noguez et al. Reference NOGUEZ, ARITA, ESCALANTE, FORNEY, GARCÍA-OLIVA and SOUZA2005, Reference NOGUEZ, ESCALANTE, FORNEY, NAVA-MENDOZA, ROSAS, SOUZA and GARCÍA-OLIVA2008). The results from Noguez et al. Reference NOGUEZ, ARITA, ESCALANTE, FORNEY, GARCÍA-OLIVA and SOUZA(2005) suggest that soil bacteria are non-ubiquitous and respond to resource distribution in this forest. The presence and activity of some species of bacterium in a given site could depend on adaptations to specific resource limitations, as shown in other studies (Belotte et al. Reference BELOTTE, CURIEN, MACLEAN and BELL2003, Cleveland et al. Reference CLEVELAND, NEMERGUT, SCHMIDT and TOWNSEND2007, Goddard & Bradford Reference GODDARD and BRADFORD2003). However, an experimental labelling approach to link unequivocally diversity and function and further studies testing resource use by specific species or functional groups of micro-organisms are still needed to understand nutrient cycling controls in tropical dry forest soils. In conclusion, our study provides a data set that suggests that C and N availability and seasonal changes in soil moisture may control microbial community dynamics, and consequently C mineralization and nitrification in these tropical deciduous forest soils.
ACKNOWLEDGEMENTS
We thank Víctor J. Jaramillo and Edmundo García-Moya for helpful discussions. We also thank the critical comments of two anonymous reviewers, who helped us improve the manuscript for publication. We would like to thank Maribel Nava-Mendoza, Tina Tønnersen and Juan Manuel Sánchez-Yañez for their technical support with chemical and microbiological analyses in the laboratory. We are also grateful to Salvador Araiza and Carlos Anaya for their help in the field work. We thank Heberto Ferreira and Alberto Valencia for assistance in data processing, and the Estación de Biología Chamela, Universidad Nacional Autónoma de México (UNAM) for logistic support during field work. This research was supported by Consejo Nacional de Ciencia y Tecnología CONACyT-Mexico (G27674-N) and Programa de Apoyo a Estudios de Posgrado-UNAM. N.M. Montaño acknowledges CONACyT (163199), UNAM and Fundación-TELMEX for PhD scholarships.
Appendix 1. List of heterotrophic bacterial species identified in the hilltop and hillslope soils of a tropical deciduous forest of Chamela, Mexico. Total relative frequencies (F %) per species at each site and date in which a species was recorded: D = dry, ER = early rainy, R = rainy seasons, as well as bacterial species richness and total common species between sites are showed. Bacterial identification was performed with the commercial Sherlock Microbial Identification System (MIDI) based on fatty acid profiling. MIDI bacteria names are based on the List of Prokaryotic names with Standing in Nomenclature (LPSN), available at the public website http://www.bacterio.cict.fr
