Soil and fertility islands
Through estimations of ANPP, a requirement to calculate GI, we could see the consequences of the grazing history in each paddock. Indeed, aerial biomass production calculated for our sites reflected fairly good the historical use contrasts we identified among plots with the qualitative survey made via interviews. We found that the non-grazed site produced more biomass than all grazed ones and that within them, the most heavily grazed site yielded far less than the others. Overgrazing affects ecosystem structure and functioning (López et al. 2011). ANPP has been proposed as a comprehensive indicator of system functioning through condensing information about energy flow and nutrient cycling (Gaitán et al. 2014). Though the assessment of ecosystem functionality was not an explicit aim of our work, our results concur with the observation of ecosystem misfunctioning at high grazing loads.
Our results evidenced the formation of fertile islands, with a differential expression between the cover categories evaluated. We could identify a significant land use effect on soil properties (Fig. 1 and table 2). Also, it was related to indicators of degraded states in grasslands as shrub cover, large interpatches, and annual grasses (Fig. 4). The effect was linked to shrubs and their surrounding bare soil spaces. Medium sand and fine sand fractions were remobilised preferentially over other fractions. Important chemical variations occurred on organic carbon and total nitrogen, two of the most important variables for soil fertility characterization. Many rangelands across the world developed fertile islands, with changes in many soil properties and vegetation arrangement (Allington and Valone 2014; Ridolfi et al. 2008). We found that bare soil areas were bigger when associated with shrub patches, compared to those associated with grasses. This resulted in enlarging the bare soil vs. patch contrast and consequently in a bigger island effect detected for shrub encroached sites. The magnitude of the heterogeneity of the resources developed in cushion dominated patches is really important. This becomes clear when comparing our reference spots with the grazed areas (Table 2). Even being the long-term less disturbed conditions, we found that soil heterogeneity between the scrubs and their paddocks was lower (but not significant) than between cushions and their interpatches.
Our quantification of island effect for subhumid Patagonian rangelands through RII agrees with previous studies on Patagonia and in the world (Aguiar and Sala 1999; Bertiller and Bisigato 1998; Gao and Carmel 2020; Okin et al. 2015; Rostagno et al. 1991). Larger barren soil spaces accompanied shrub cover growth. Shrub patches of the cushion plant M. spinosum and their associated bare interpatches exhibited more heterogeneity than grasses. Recent studies found an important influence of canopy size on the fertility island effect (Zhao and An 2021). Ding and Eldridge (2021) found a greater effect in shrubs than grasses for both biotic and abiotic variables, also with increments positively related to aridity. Howard et al. (2012) found microclimatic amelioration under shrubs, for nutrients cycling, soil properties and diversity indices. Cushion plants are a common life form in harsh environments across the world (Aubert et al. 2014). Cushions are an important fertile island formation agent since their canopies are good in trapping and maintaining sediments beneath them and in producing facilitative interactions (Cavieres and Badano 2009; Gavini et al. 2019). Mihoč et al. (2016) measured nurse effects along an environmental gradient, finding higher fertility and activity under cushion shrubs than any other life form.
We found that soil particles intervening in remobilisation (FS, MS, CLAY and SILT) took intermediate values in the soil of reference spots, compared to the soil beneath shrub and grass canopies and barren areas (Fig. 2). Plants with different shapes and traits don’t behave as sediment traps equally (Cavieres and Badano 2009; Mihoč et al. 2016). Tall shrubs with Y-shaped crowns have less capacity to capture and retain airborne sediments than those with hemispheroidal-canopies (Li et al. 2007). Besides, the larger size of the reference spots and the greater distance from bare soil spaces, in comparison to the other vegetated patches, also reduces the accumulation of material from bare areas (Ravi et al. 2011). We suppose the differential trapping ability together with the greater distance from redistributed particles sources contributed to maintaining less altered particles distribution of our reference spots compared to areas dominated by cushion shrubs and grasses. Our results showed that soil of grass and cushion shrub patches differ to a large degree from reference spots in the particles intervening in remobilisation (Fig. 2). The higher content of organic matter could also mediate this effect (Figs. 1 and 2). The organic matter reduces soil susceptibility to erosion by enhancing the stability of its aggregates, which turns into lesser particle mobilisation (Cambardella and Elliott 1993; La Manna et al. 2018). Thus, because of less particle entrapment and also to fewer losses, we assume that the soil of these spots exhibits a situation closer to the existent before intense disturbance from sheep grazing.
Our results showed preferential remobilisation of soil, with differential effects at deposition patches (sink) and source areas. Sand medium and fine particles were higher in the soil beneath canopies. In source patches (bare soil spaces), we found consequent increments of fine particles (clay and silt) and gravel. In Patagonia, these redistribution effects were found for climatic regimes ranging from arid to subhumid (Kröpfl et al. 2013; Paruelo and Golluscio 1993; Rostagno 1989) and other rangeland systems of the world (Li et al. 2007; Navas et al. 2017; Zhang et al. 2011). Volcanic soils are wind-erosion-prone, particularly in rangelands, because of high proportions of volcanic glass in their sand fraction, strongly vesicular or pumiceous (McDaniel et al. 2012). This turns these particles lighter than sand particles from non-volcanic soils (Nanzyo and Kanno 2018). Recent water erosion studies in volcanic soils in the same region of Patagonia found remobilisation of coarse and very coarse single sand particles in the rangelands, while erosion processes in afforested areas involved the removal of microaggregates rich in organic matter and silt fractions (La Manna et al. 2021). Based on our results and the spotted vegetation pattern we found at our study sites (Aguiar and Sala 1999), we suppose remobilisation processes taking place are mainly wind-driven. The erodible fraction was lower and gravel content was higher in barren soil spaces related to shrub patches. Thus, soil losses effectively happened with more intensity there, while vegetated patches showed the opposite situation.
The soils we sampled under plant canopies had better chemical properties than those in barren areas. This was an expected finding since many other studies reported it for this region and other rangelands of Patagonia and the world (Aguiar and Sala 1999; Cheng et al. 2004; Hao et al. 2016; Kröpfl et al. 2013; Mihoč et al. 2016; Rostagno et al. 1991). Nitrogen and organic matter are within the most cited chemical properties with increases (Kröpfl et al. 2013; Rostagno 1989; Rostagno and del Valle 1988; Zhao and An 2021). Other studies reported increasing values for electric conductivity and decreasing pH beneath vegetation canopies (Rostagno et al. 1991; Zhao and An 2021). Indeed, a review has reported a general lowering pH value with an increase in shrub dominance (Eldridge et al. 2011). As for textural properties, soil fertility indicators were more heterogeneous for shrubs than for grasses.
The soils of our reference spots (scrub thickets) were far more fertile (OM and N) than the soil of grass and shrub patches. Differential C and N turnover rates due to litter quality can explain these variations. Many studies focused on woody species leaf traits found that deciduous plant litter mineralises faster than evergreens (Carrera et al. 2005; Cornwell et al. 2008; Satti et al. 2003). Evergreens long-lasting tissues usually contain higher levels of secondary compounds in leaves (Bertiller et al. 2005; Carrera and Bertiller 2010; Carrera et al. 2005, 2008; Mazzarino et al. 1998; Saraví Cisneros et al. 2013). Most shrubs, both in our reference spots and in encroached areas, have short-lived foliage. Despite this, the litter quality framework is worth value. Our shrub patches comprised deciduous M. spinosum cushions. This shrub has a high concentration of secondary metabolites in leaves (Cavagnaro et al. 2003), seeds (Folgarait and Sala 2002), flowers and fruits (Seoane et al. 2011). Also, it yields a moderate quantity of essential oils (Guerra et al. 2012). However, scrubs species of the reference spots were mainly of Discaria and Colletia genus, tribe Colletieae (Rhamnaceae). Their low-strengthen deciduous leaves (De Paz et al. 2013) don’t have detectable essential oils (Guerra et al. 2012). These chemical and physical features turn into very different decomposition rates (k). Under similar conditions, M. spinosum litter k values ranged from 0.3 to 0.38 (Araujo and Austin 2015), while k was as high as 1.17 for D. articulata (de Paz et al. 2017). Moreover, sunlight is an important control of litter turnover for environments where vegetation cover is heterogeneous or scarce (Adair et al. 2017; Araujo and Austin 2015; Bosco et al. 2016). It speeds up litter degradation as much as twice through photodegradation and photofacilitation (Berenstecher et al. 2020). The growth habit of M. spinosum implies the formation of a compact canopy with structural avoidance of excessive irradiance (Damascos et al. 2008). Conversely, vertical stems of shrubs or trees from tribe Colletieae (Tortosa et al. 1996) enables sunlight to reach the basement. We propose that soil fertility under high shrubs reflects the light-enhanced degradation of the high-quality litter produced by these plants.
We didn’t find extensive indications of amorphous materials at our study sites. We expected to have these minerals, in the same line of previous assessments for soils with similar moisture regimes in this region (Broquen et al. 2005; La Manna et al. 2020). The sodium fluoride pH test only detected allophane or imogolite consistent values (Irisarri 2000) in one of our study sites (AN) and one of our scrub references samples. Hence, we assumed the non-general current presence of these non-crystalline clays. The hilly landscape of this area determines a wide variety of positions and expositions that enable the coexistence of Andisols and Andic Mollisols (Irisarri et al. 1995). In previous studies on the same sites, we found NaF pH imogolite consistent values in one sub-superficial (25 cm depth) sample (Vogel and La Manna 2018). In our present assessment, we only found imogolite NaF pH values in one of the non-disturbed tall shrub spots and the grassy area excluded from livestock grazing. Except for these scattered spots, our study sites and most of the region had historically high levels of livestock grazing. Such kind of disturbance leads to important losses of vegetation cover, exposing superficial layers to direct climate action. Intense disturbance by human activities can lead to the transformation of non-crystalline clays into halloysite-type minerals because of continuous exposure to desiccation and re-wetting of aggregates (Hernández et al. 2012). Other assessments in the region suggested this transformation (La Manna et al. 2018), which is irreversible and relatively fast (McDaniel et al. 2012). The evaluated sites faced strong erosion processes, as we showed before through the analysis of soil particles redistribution, fertility island effect and soil profiles. If imogolite developed in these volcanic ash-originated soils, its irreversible transformation to halloysite may be possible.
Vegetation changes
We found a clear association of higher levels of use intensity with changes in community composition and vegetation structure. One of the most remarkable by its consequences on secondary productivity is the cover reduction of perennial grasses. Previous studies informed this for the sub-Andean district (Aguiar et al. 1996; Anchorena and Cingolani 2002; Bertiller and Bisigato 1998; Cesa and Paruelo 2011; Gaitán et al. 2018; Paruelo et al. 2004, 2008) despite evidence of a not universal association (Lezama et al. 2014). Grazing disturbance favour grasslands to shrublands transition (Kröpfl et al. 2013; Ridolfi et al. 2008). Besides a grass cover contraction, we also found an increment in total shrub cover with increases in disturbance (grazing) intensity. The most relevant species (>1% total frequency) found in sites with high grass cover were Pappostipa speciosa (Pap.spe) and Bromus stamineus (Bro.sta). On the opposite edge, where grass cover reduced, Mulinum spinosum (MUL.spi) and Festuca pallescens (Fes.pal) were the most notorious species (Table 3 and fig. 4). Increments in cushion shrubs were more associated with grazing intensity and bare soil than total shrubs, which shows their encroacher trait in disturbed sub-Andean rangelands (León and Aguiar 1985). Grazing is also a major driver of fertility island formation (Cai et al. 2020; Allington and Valone 2014). As discussed before, we found a greater fertility island effect associated with shrub patches and their paired large bare soil interpatches (Fig. 1 and Table 2). Moreover, we also verified a determinant effect related to increments of shrub in community contribution and cover (Fig. 4).
We detected that larger bare interpatches accompanied the increase in total shrub cover (fig. 4). Lin et al. (2010) proposed patch edge as a key indicator of water and sediments trapping efficiency. Larger patches like those formed by shrubs have less edge by area than smaller patches like those of grasses and forbs. Our survey results support this asseveration since grass patches were higher in sand remobilised particles, nitrogen and organic matter than those of shrubs (Fig. 2). Also, the bare interspace between shrubs was bigger (Fig. 4) and more heterogeneous with their corresponding patch than those of grasses (Fig. 1, Table 2). These increased losses in shrub interspace areas together with soil amelioration under both grass and shrub canopies created the island effect we found.
We also found that the percentage of uncovered area incremented with growing disturbance intensity (Fig. 4). Despite not being an effect unrestrictedly detected (Cesa and Paruelo 2011; Eldridge et al. 2011), this response of rangelands to grazing has received consensus both for Patagonia and other ranges (Aguiar et al. 1996; Cheng et al. 2004; Van Auken 2000; Verón and Paruelo 2010). Grazing and trampling by herbivores change the topsoil conditions of interpatches, playing a key role in the fertility island effect (Allington and Valone 2014) and even desertification (Schlesinger et al. 1990). Indeed, sheep grazing habits may magnify these effects of trampling (Tóth et al. 2018). Our results verified the description of the present range status we made in the methods section. There, we emphasised the visually stunning experience of large barren spaces related to the domination of shrubs and high historical disturbance pressure in almost the entire area.
Our assessment detected structural changes, such as increments of forbs and annuals (in cover and proportional contribution to community) (Fig. 4). Literature mentions them as indicators of grazing disturbance for sub-Andean rangelands (Anchorena and Cingolani 2002; Cesa and Paruelo 2011; Paruelo and Golluscio 1993). These changes were both associated with grazing pressure increases but were not much related to each other. Community with forb increments had Poa ligularis (POA.lig), Acaena pinnatifida (ACA.pin) and Carex argentina (CAR.arg) as the most relevant species (>1% total frequency). The most relevant in communities where annuals increased was Vulpia sp (VUL.sp) (Table 3 and Fig. 4). When a patch collapse after heavy disturbance, the resources flow to bare spaces promoting recovery processes in rangelands (Kröpfl et al. 2013). Resources are lost if surrounding patches and interpatches cannot catch them efficiently. Following Lin et al. (2010) we assume that having less proportional patch edge, shrub patches trap resources depleted from bare soil with lesser efficiency than grasses. Thereby, these encroached ranges reached soil conditions limiting vegetation growth. In that situation, some species can take advantage of spatial heterogeneity created by disturbance.
At our study sites, annual species accompanied shrubs cover growth, an effect previously reported for moderate and heavily grazed rangelands (Noy-Meir et al. 1989). Annuals increment usually follows perennials decrease (Milchunas and Lauenroth 1993). These species prevail in bare interpatches because of advantages such as tolerance to harsh environmental conditions, higher seed production, shorter and prostrate canopy, secondary compounds and life cycles suitable to benefit from any benign microclimatic condition (Anchorena and Cingolani 2002; Gao and Carmel 2020; Milton et al. 1994; Noy-Meir et al. 1989). We verified clay increments in superficial soil of bare shrub interpatches (Fig. 2). This causes a xerification effect in Subandean highly eroded soils, diminishing water available for plants uptake (Paruelo and Golluscio 1993). The situation described here is comparable to the shrub-steppe of NW Chubut named state IV, with a very unlikely return to a more productive state because of irreversible soil changes (Paruelo and Golluscio 1993). In such harsh conditions, reinforced by a high content of coarse fragments (Table 2 and fig. 3), we assumed as possible the constitution of niche gaps in the highly eroded big barren interspaces, where only annual grasses could complete an entire life cycle. Annuals are the most successful trait for such levels of resources depletion and spatial heterogeneity.
We found forbs gained participation in cover, being most of them perennial and native. We didn’t observe the increment in exotics with grazing intensity reported for other subhumid rangelands (Lyseng et al. 2018). Many forbs share key traits with successful exotic invader plants, like grazing tolerance/avoiding. Forbs are usually smaller than shrubs and grasses, which enables them to explore microhabitats created by disturbing activities such as herbivory, increasing plant richness (Jobbágy et al. 1996). Vegetation cover may become more fragmented because of destructive grazing and trampling (Lin et al. 2010), creating spatial heterogeneity and opening niche gaps, thus encouraging species coexistence (Gao and Carmel 2020). We believe that grazing intensity could lay behind the independent response of forbs to abiotic heterogeneity reported by Jobbágy et al. (1996). We observed vegetation layers superposition together with increasing forbs cover (Fig. 4) in our study sites, variables which are showing the mentioned niche heterogeneity. Indeed, in those sites with less bare soil and shrub cover, and smaller barren spaces, growth several life forms of perennial vegetation other than shrubs, like native forbs (Fig. 4). This situation pairs with grass and grass-shrub steppe of SW Chubut named as states II and III by Bertiller and Defossé (1993). Maintaining current grazing pressure will cause a probable transition to shrub-grass steppes of M. spinosum or Acaena spp. We consider that grazing management should focus on sites like these since they exhibit less soil heterogeneity and still sustain life forms other than shrubs and annual grasses.
Our results showed that increasing grazing intensity exerted a positive effect on total richness. This effect manifests along all grazing intensities evaluated with exception of the highest, were richness finally declines (Fig 5). While shrubs and grasses responded by replacement in species of one life form by the other, perennial forbs (and secondarily annual grasses and forbs) represented an addition to the community (Fig. 4). Golluscio and Mercau (1995) identified shrubs and cushions as increasers, some grasses and palatable forbs as decreasers and other grasses and forbs with maximum cover values at mid grazing intensities. Grazing induces changes in interspecific competition, producing a shift in species abundance that otherwise won’t succeed because of resistance of non-grazed grasslands to invasion (Lyseng et al. 2018). The effects of grazing on grasslands richness have been controversial, and attempts to generalise them had arrived at different conclusions (Eldridge et al. 2011, 2016; Herrero-Jáuregui and Oesterheld 2018; Howard et al. 2012). For Subandean grasslands, while some authors didn’t observe major changes (Anchorena and Cingolani 2002; León and Aguiar 1985) most studies negatively related richness and grazing intensity (Cesa and Paruelo 2011; Golluscio et al. 1982; Lezama et al. 2014; Paruelo et al. 2004). We found a hint for our controversial results in the warning given by Bertiller and Bisigato (1998) about the drastic magnitude of changes in the more humid grasslands of Patagonia.
We verified via the Global Aridity Index dataset (Trabucco and Zomer 2018) that most of the Subandean vegetation assessments take place at the eastern semi-arid portion, while western subhumid zones were less surveyed. Milchunas and Lauenroth (1993) reviewed moisture-dependant effects of grazing on diversity, with increases for subhumid rangelands at low-moderate to moderate intensities. They postulated that grazing influences species composition more in subhumid than in arid environments because vegetation removal enhances the intensity of plant interactions. A recent global meta-analysis concluded that grazing increases plant richness, particularly for dry-subhumid or wetter grasslands with an aridity index (AI) > 0.51 (Gao and Carmel 2020). Destructive animal activity (trampling, grazing) over edible and highly competitive species may release resources such as space, light, water and nutrients for those with grazing avoidance strategies. Also, non-destructive activities (urine and dung deposition) may reinforce this vegetation heterogeneity and increase richness over a threshold of moisture limitations for plant growth (AI 0.51) (Gao and Carmel 2020). Grazing produced modest increases in richness in mesic sites of north-American temperate grasslands (Lyseng et al. 2018) with structural and functional convergence with Patagonia (Paruelo et al. 1998b). However, once surpassed a threshold, heavily disturbed sites would present a downslope in the curve of diversity changes (Milchunas and Lauenroth 1993). Effectively, the highest grazing intensities exhibited less richness of all our study sites in our assessment (fig. 5), in line with previous reports for sub-Andean district (Cesa and Paruelo 2011; Golluscio et al. 1982; Lezama et al. 2014; Paruelo et al. 2004). Notwithstanding this, moderate grazing intensities at the subhumid sites we evaluated created niche gaps that favoured the coexistence of grazing tolerant and intolerant plant species and resulted in larger richness mainly because of perennial forbs species addition.