Influence of woody species on aerial growth of perennial grasses in semi-arid rangelands of central Argentina

1 Departamento de Agronomía, Universidad Nacional del Sur, 8000 Bahía Blanca, Argentina. 2 Consejo de Investigaciones Científicas y Técnicas (CONICET), Argentina. 3 Centro de Recursos Naturales Renovables de la Zona Semiárida (CERZOS), Argentina. 4 Comisión de Investigaciones Científicas de la Provincia de Buenos Aires (CIC), Argentina. Address Correspondence to: Francisco R. Blazquez, e-mail: francisco.blazquez@uns.edu.ar Recibido / Received 4.V.2014. Aceptado / Accepted 30.V.2014 Resumen. La interacción entre especies leñosas y gramíneas en pastizales semiáridos del centro de Argentina fue evaluada a través del impacto de las especies leñosas sobre el crecimiento de las gramíneas durante un período de dos años. El estudio involucró a dos especies leñosas de diferentes características morfo-fisiológicas y a dos gramíneas perennes de ciclo otoño-inverno-primaveral altamente preferidas por el ganado doméstico. Se determinó la longitud total verde (LTV) y número de verdes hojas (NHV) por macolla en plantas de las gramíneas creciendo bajo la canopia de las especies leñosas y en sitios abiertos del pastizal; mientras que, la temperatura del suelo, pH y contenido de agua, nitrógeno, fósforo y materia orgánica se estimaron en los mismos sitios. También se registró la precipitación durante el período de estudio. Las gramíneas que crecieron bajo la canopia de las especies leñosas tuvieron mayor LTV cuando la precipitación fue similar o superior al promedio histórico del sitio de estudio; sin embargo, no se detectaron diferencias en el NHV. El contenido de nitrógeno y de materia orgánica en el suelo debajo de la canopia de las especies leñosas fue mayor que en los sitios abiertos; mientras que, las diferencias en los otros parámetros del suelo fueron escasas y las posibles causas de ello son discutidas. En general, los resultados del presente estudio sugieren que las especies leñosas ejercen un efecto facilitador sobre las gramíneas perennes en los pastizales semiáridos del centro de Argentina.


INTRODUCTION
Grasslands, shrublands, savannas, forests and drylands, which can be collectively addressed as rangelands, occupy about 45-52% of the Earth's surface (Matthews, 1983).Plant communities in these ecosystems are typically formed by a continuous layer of grass species and a discontinuous one of woody species.The influence of herbivory, fire, soil and climate on both vegetation layers have been thoroughly studied (Sala, 1988;McNaughton, 1991).However, the understanding of the interactions between woody and perennial grass species in semi-arid ecosystems still remains elusive.
The coexistence of woody and perennial grass species in semi-arid ecosystems has been attributed to the combined effect of water and nutrient availability, fire frequency and intensity, and herbivory (Werner, 1990).The structure of the community is shaped by the interaction among these biotic and abiotic factors, and the complex balance of competitive and facilitative mechanisms that operate simultaneously between neighbouring species (Callaway & Walker 1997).
It has been proposed, through the 'stress-gradient hypothesis' (SGH), that the relative frequency of competitive and facilitative mechanisms varies inversely across abiotic stress gradients (Bertness & Callaway, 1994).According to this hypothesis, facilitation is more common under stressful abiotic conditions, even though it may decrease or cease under extreme stress, while competition prevails under more favourable abiotic conditions.It has also been suggested that the balance between these mechanisms depends, to a large extent, on the characteristics of the species involved in the interaction (Choler et al., 2001).
The aftermath of the interaction between woody and perennial grass species has been assessed by comparing the grass species productivity at sites located under the canopy of woody species individuals with that in open rangeland areas.Several mechanisms have been identified as responsible for the differences found in grass productivity between sites.Competition for light, water and nutrients has been the major recognized mechanism (Walter, 1971;Walker & Noy-Meir, 1982;McMurtrie & Wolf, 1983;Scholes & Archer, 1997).Thereafter, the most common facilitative mechanisms identified have been the (1) increase in soil nutrient content and structure beneath the canopy of woody species, accounted for by litter inputs from the woody species (Whitford et al., 1996;Weltzen & Coughenour, 1990), and (2) amelioration of grass water status, through a reduction in the amount of incident solar radiation which translates into lower temperatures and, ultimately, into reduced evapotranspiration beneath the woody species canopy (Ludwig et al., 2004).
In semi-arid rangelands of the southern Caldenal (Cabrera, 1976), located in central Argentina, land is almost exclusively used for cattle raising.Long-term overgrazing and the lengthening of the fire-free period have induced changes in the structure of plant communities.This has been through an increase in abundance of woody species and a decrease in the abundance and productivity of perennial grass species (Distel & Bóo, 1996).Woody species control and management strategies aimed to restore the grassland productivity have been elaborated.However, the extent to which woody species should be removed, in terms of rangeland sustainability, is yet unknown.Hence, studies attempting to elucidate the effect of woody species on growth of perennial grass species in this region are needed.
The specific objective of this study was to evaluate the effect of two morpho-physiologically different woody species on growth of two cool-season perennial grass species highly preferred by domestic livestock in the southern Caldenal.This was achieved after assessing of the impact of those woody species on grass and soil parameters that are major determinants of grass growth.

MATERIALS AND METHODS
Study site.This study was conducted at a representative site of the Caldenal, located in the south-eastern corner of the province of La Pampa in central Argentina (38˚ 45' S, 63˚ 45' W).The leveled, 20-ha study site was (1) located within a pasture with no recent fire history and (2) fenced since 1982 to exclude livestock grazing.
The climate of the region is temperate and semiarid (Instituto Nacional de Tecnología Agropecuaria, 1980).Mean monthly air temperatures range from a low of 7 ˚C in July to a high of 24 ˚C in January, with an annual mean of 15 ˚C.Mean annual rainfall is 400 mm, highly variable between years and with peaks in autumn and spring.Mean annual potential evapotranspiration is 800 mm.Annual precipitation during the study period was 369 mm in 2011 and 510 mm in 2012 (Servicio Meteorológico Nacional, 2013).Its distribution was approximately similar to that of the long-term average (40-y), except for deeper water deficits in summer and winter, and the higher rainfall records in spring (Fig. 1).Soils are medium to coarse-textured Calciustolls (Sánchez & Lazzari, 1999), with a petrocalcic horizon commonly found at 60-80 cm depth.
Influence of woody species on aerial growth of perennial grasses Experimental design.The species chosen for the study were P. caldenia and L. divaricata, a warm-season deciduous and an evergreen woody species, respectively; and N. clarazii and P. napostaense as representatives of perennial grasses highly preferred by livestock (Bontti et al., 1999;Bóo et al., 1993;Pisani et al., 2000).Six plants of each woody species were randomly selected at the study site.An individual of each grass species was chosen both beneath the canopy of each woody species individual and in the open area (~2 m away from the shrub crown).Six tillers were labelled in each grass plant (3 in the centre and 3 in the outer boundary of the crown) using coloured wire rings.
Aerial growth measurements.Two morphogenetic traits were measured on each tiller in order to assess the aerial growth of grasses; namely the total green length (TGL), and the number of green leaves (NGL).Measurements were made periodically during two growing seasons (May-Dec 2011 and March-Dec 2012) with an increasing frequency towards the end of the growth cycle (i.e., from 30 to 15-d interval between measurements).At the beginning of each growth cycle plants were defoliated to almost ground level (4 cm).Afterwards, on each tiller, the TGL was calculated as the sum of the following measurements made using a 1-mm graduated ruler: (i) green blade length, measured from the tip of fully expanded leaves to their own ligule, or from the tip of growing leaves to the ligule of the previously fully expanded leaf, (ii) pseudostem measured from the soil to the ligule of the last fully expanded leaf, and (iii) true stem measured from the ligule of the flag leaf to the base of the panicle.The NGL was recorded on each date by counting the number of growing or fully expanded leaves that did not have signs of senescence.

Soil sampling.
At the time of each aerial growth measurement, soil temperature (5 cm deep) was recorded by using thermocouples type E (chromel-constantan) connected to a  microvoltmeter (HR-33T, Wescor Inc., Logan, UT, USA).Measurements were made both beneath the canopy and in the open area (~2 m away from the shrub crown) of each woody species individual.Simultaneously, in 2011, one soil core (3 cm diameter x 20 cm deep) was collected beneath the canopy of each woody species while another six soil cores were taken in the open area (>2 m away from any woody species canopy).Samples were taken to the laboratory to determine soil gravimetric water content.In 2012, the soil sampling procedure was modified as follows: one soil core was taken beneath the canopy, while another one was taken in the open area (~2 m away from the shrub crown) of each woody species individual.
Statistical analyses.Data were analysed as a split-split plot design with (1) woody species individuals as plots, ( 2) locations (beneath canopy vs. open area) as subplots, and (3) sampling date as sub-subplot for each grass species and year separately.The model effects were woody species, location and perennial grass species.Data of TGL were log-transformed to meet normality and homocedasticity.Means were compared by Tukey HSD test (Snedecor & Cochran, 1980) when significant differences were found.All statistical analyses were performed with JMP 7.0 (SAS Institute 2007).

RESULTS
Overall, the TGL and NGL of both grass species varied across time both within each year and between years (p<0.05).Both growth parameters showed their lowest values (p<0.05) during autumn and winter in 2011 and 2012, and increased significantly in spring (Fig. 2 and 3).There was no significant effect of location on the NGL (p<0.05).However, the TGL of grasses was higher beneath the canopy of the woody species than in the open areas during spring 2011 for N. clarazii, and at most of the sampling dates in 2012 for both grass species (Fig. 2).Generally, there was no significant effect of the woody species on TGL or NGL.The only exceptions were N. clarazii for TGL in late spring 2011, and both grass species in late spring 2012: TGL of grass plants located beneath the canopy of P. caldenia was higher than that beneath the canopy of L. divaricata (p<0.05) (Fig. 2).
Soil temperature was similar (p>0.05)among woody species and location during autumn and winter both in 2011 and 2012.Differences in soil temperature were mostly apparent in spring in both years (Table 1

Total green length (cm)
the open areas at most of the sampling dates both in 2011 and 2012; however, there were no differences (p>0.05) between locations in L. divaricata (Table 1).
Soil water content was similar (p>0.05)among woody species and location throughout 2011 (Table 2).The only exception occurred at the last sampling date, when soil water content was significantly lower beneath the canopy of P. caldenia than beneath that of L. divaricata (Table 2).In 2012, there was no significant (p>0.05)effect of location on soil water  content.The only exception was L. divaricata (p<0.05) at the first sampling date.However, there were significant differences (p<0.05) in soil water content between woody species at four sampling dates (Table 2).
Soil N and OM content significantly differed (p<0.05) between locations regardless of woody species, values were higher beneath each woody species canopy than in the open areas (Table 3).On the other hand, there were no significant differences in P content or pH (p>0.05).
Influence of woody species on aerial growth of perennial grasses

DISCUSSION
Overall, woody species effect on grass TGL was neutral in 2011 while positive in 2012.This difference was probably accounted for by differences in precipitation between years.Accordingly, in 2011, there were no significant differences in TGL between N. clarazii and P. napostaense both beneath the canopy of the woody species and in the open areas during autumn and winter (from May to September).TGL of both grass species was low (less than 5 cm) during that period (Fig. 2), when rainfall records were below the long-term average (Fig. 1).From this time onwards, however, rainfall was above the long-term average until the end of this growth cycle.This may have allowed both grass species to resume growth immediately.Together with the onset of flowering, this might help explain the increase in TGL towards the end of the growth period.This is because during the reproductive phase, grass internodes elongate protruding first the leaves that were formerly within the pseudostem, and immediately afterwards, the panicle (Langer, Blazquez FR et al., FYTON 83 (2014) Table 1.Soil temperature at 5 cm soil depth beneath the canopy and in the open areas (~2 m away from the crown) of P. caldenia and L. divaricata in 2011 and 2012.Values are means of 6 replicates; SEM equalled 0.8 both in 2011 and 2012.Within each year and column, values followed by the same letter are not significantly different (p>0.05).

Soil temperature (°C) 2011
May  ).These leaves, as well as the peduncle of the panicle, added up to the TGL and would account for the marked increase of this parameter in spring.Notwithstanding these differences in TGL across time, differences between locations in spring (from October to December) were only found in N. clarazii for both study woody species (Fig. 2).This might be attributed to species-specific characteristics: it has been shown that N. clarazii has a higher competitive ability than P. napostaense (Saint Pierre & Busso, 2006).Thereafter, a higher resource acquisition ability might explain the faster response to increased water availability in N. clarazii than in P. napostaense (Fig. 2).
In 2012, TGL in N. clarazii and P. napostaense was higher beneath the canopy of either P. caldenia or L. divaricata than in the open areas throughout the growth cycle (Fig. 2); this is in agreement with near-or above-average monthly rainfall records (Fig. 1).The differences in TGL between locations cannot be attributed to differences in soil water content since these were scarce (Table 2).They rather might be explained by differences in soil N and OM content, since these parameters where higher beneath woody species than in open areas (Table 3).Similar results were reported by Belsky et al. (1989) for Acacia tortilis and Adansonia digitata in African savannas, and by López-Pintor et al. (2006) for Retama sphaerocarpa in Mediterranean grasslands.López-Pintor et al. (2006) found no difference in soil P content and pH with distance from the shrub crown.Belsky et al. (1989) found no differences in soil pH with distance from tree trunks, but a higher soil P content immediately next to the tree trunk; nevertheless, soil P content declined sharply within the canopy area (Belsky et al., 1989).Overall the results of our study were in agreement with these findings as soil P content and pH were similar between locations.
Grasses growing beneath the canopy of P. caldenia tended to show higher TGL than those beneath the canopy of L. divaricata.However, significant differences were only found in late spring 2011 for N. clarazii, and in late spring 2012 for both grass species (Fig. 2).This was coincident with lower soil temperatures in spring beneath the canopy of P. caldenia (Table 1).These results are in agreement with findings of Buschiazzo et al. (2004).These authors showed that while soil temperatures were ~35 ˚C beneath the canopy of P. caldenia during summer in the Northern Caldenal, they could reach ~65 ˚C in open areas.Differences in soil temperatures between locations for P. caldenia, but not for L. divaricata, may probably be attributed to differences in canopy architecture.Differences in plant architecture might account for differences in the amount of solar radiation reaching the ground, and hence, for differences in soil temperature.Prosopis caldenia exhibits a hemispherical canopy inclined towards the cardinal point from which the highest amount of solar radiation is received during this species growth cycle (i.e., from the south during spring and summer).However, there seems to be no specific pattern in the canopy architecture of L. divaricata, which shows an inverted cone shape (Cano, 1988;Kröpfl et al., 2012).
Differences in soil temperatures should be expected to turn into differences in soil water content as a result of influencing the evapotranspiration rates (Smith et al., 1994).Accordingly, soil water content has been found to be higher beneath the canopy of woody species than in open areas in the Caldenal (Llorens & Frank, 1999).However, in the present study, soil water content did not differ between the study locations (Tables 1  and 2).The only exceptions were the lower temperatures recorded beneath the canopy of P. caldenia in spring.This may be explained by a higher water demand from grasses due to their increased growth rate during the flowering period (Langer, 1972).This may have overridden any increase in soil water content accounted for by the shade provided by the woody species canopy.Our results are in partial agreement with those of Belsky et al. (1993).These authors, in a similar study in the African savannas, found that soil water content was higher beneath the canopy of trees than in open areas due to the lower temperatures and reduced evapotranspiration rates at the beginning of the grass growth cycle.However, with the progress of the growing season, grasses grew faster beneath the canopy of trees than in open areas, extracting increasingly greater amounts of soil water and, thereby, equalling soil water content between locations by the end of the grass growth cycle.
The lack of differences in soil water content between locations (in spite of differences in soil temperatures) can also be explained by the counterbalancing effect of the interception of precipitation by the shrub canopies during small rainfall events.Accordingly, Kröpfl et al. (2002) reported that L. divaricata often significantly reduced soil water availability beneath its canopy.Belsky et al. (1989) showed that the proportion of precipitation intercepted by the canopy of trees was related to the magnitude of the rainfall event, ranging from 100 to 0% for rainfall events from 2 to 20 mm, respectively.Another possible explanation for the higher TGL of grass species beneath the canopy of P. caldenia and not of that of L. divaricata may be related to the differences found in soil N and OM content.Although not statistically significant, both soil parameters tended to be higher beneath the canopy of P. caldenia than that of L. divaricata (Table 3).This can be attributed to woody species differences in morphology and leaf habit.As discussed by Whitford et al. (1996), differences in canopy architecture (i.e., hemispherical vs. conical) result in differences in wind turbulence and, hence, in fine particulate and litter accumulation.These authors found that the accumulation and retention of litter was higher beneath the canopy of hemispherical than conical shrubs.Although wind turbulence and amount of litter deposition were not measured in the present study, the latter might provide an explanation for the higher N and OM content in soils beneath the hemispherical canopy of P. caldenia than beneath the conical canopy of L. divaricata.In addition, leaf habit (i.e., deciduous vs. evergreen species) can also affect soil N and OM content as a result of differential deposition of litter.It has been shown that leaf longevity Blazquez FR et al., FYTON 83 (2014) in deciduous trees in temperate forest range from 90 to 180 days, while leaf longevity of evergreens in temperate forests are considerably longer, often more than 500 days (Kikuzawa, 1995).The longer leaf life span on evergreen species plays an important role as a nutrient conservation mechanism, in contrast to the high resorption efficiency typical of deciduous species (Aerts, 1996).It is not yet known, however, the extent of differences in the recycling of nutrients between deciduous and evergreen woody species in the study ecosystem.
Even though incident solar radiation was not assessed in the present study, its effect on grass growth parameters cannot be overlooked.It has been demonstrated that a decrease in light intensity causes an increase in leaf length and a decrease in leaf dry weight (Langer, 1972).Data of specific leaf area (i.e., the ratio of leaf area to dry mass) should be the subject of future research to assess the extent to which grass TGL could be used as a predictor of grass productivity in the study ecosystem.
The other growth parameter evaluated in the present study (NGL) varied across time within a year and between years, but it was not affected either by woody species or by location for both grass species (Fig. 3).Overall, grass tillers hold a relatively constant number of green leaves during their growth cycle (Davies, 1977); nevertheless, NGL during the reproductive phase tended to be higher than that in the vegetative phase as a result of internode elongation.In addition, the NGL can be modified by water and nutrient availability (Longnecker & Robson, 1994).According to this, differences in NGL across time within a year and between years might be the outcome of differences in soil water content and the onset of flowering, as it was explained for TGL.On the other hand, the lack of differences in the NGL between locations can be partially attributed to their similar water content.Likewise, the differences in N and OM content between locations might have not been sufficient to trigger differences in the NGL between the study grass species.
The results of the present study are in agreement with the stress-gradient hypothesis.According to it, facilitative mechanisms would prevail over competitive ones in a semi-arid ecosystem like the Caldenal.This is because the stress imposed by the abiotic environment does not reach a threshold over which facilitation ceases.Our results contribute to the understanding of the interaction between woody and grass species.This is an essential preliminary step towards the design of woody species management strategies aimed to improve grassland productivity in semi-arid rangelands of central Argentina.

Fig. 2 .
Fig. 2. Mean total green length (TGL) on tillers of N. clarazii and P. napostaense beneath the canopy and in open areas (~2 m away from the crown) of P. caldenia and L. divaricata in 2011 and 2012.Values are means of 6 replicates.Note graphic scale change in y axis for P. napostaense in 2011 and 2012.
Fig. 3. Mean number of green leaves (NGL) on tillers of N. clarazii and P. napostaense beneath the canopy and in open areas (~2 m away from the crown) of P. caldenia and L. divaricata in 2011 and 2012.Values are means of 6 replicates.Note graphic scale change in y axis for N. clarazii in 2011 and 2012.

Table 2 .
Soil gravimetric water content at 0-20 cm soil depth beneath the canopy versus open areas (~2 m away from the crown) of P. caldenia and L. divaricata in 2011 and 2012.Values are means of 6 replicates, SEM equalled 0.6 in 2011 and 0.7 in 2012.Within each year and column, values followed by the same letter are not significantly different (p>0.05).

Table 3 .
Soil nitrogen, phosphorus, organic matter content and pH at 0-20 cm soil depth beneath the canopy versus open areas (~2 m away from the crown) of P. caldenia and L. divaricata in March 2012.Values are means of 6 replicates ± SD.Within each column, values followed by the same letter are not significantly different (p>0.05).