Introduction
Drylands (arid, semiarid, and dry subhu mid lands) cover approximately 46% of the Earth’s surface and are home to more than 3 billion people (Mirzabaev et al., 2019). They are also home to surprisingly vulnerable ecosystems sensitive to climate chaos, soil and plant cover loss, and other widespread global changes leading to eco system degradation and human misery (Huang et al., 2016). Moreover, most dry lands worldwide are demonstrably losing biodiversity and topsoil at a catastrophic rate, as a direct result of poor steward ship and anthropogenic climate change (Cherlet et al., 2018). This is resulting in increased poverty, food insecurity, and disease burden in rural communities that depend more or less entirely upon their local ecosystems for sustenance and sur vival (Mirzabaev et al., 2019). Thus, dry lands, and the people who inhabit them, urgently need large-scale ecological resto ration and rehabilitation.
Shrubs are often the dominant life-form in the vegetation communities of many drylands, and, consequently, rep resent key elements or candidate ‘dry land framework species’ (DFS) in kick starting effective, long-lasting ecological restoration or rehabilitation. However, an overwhelming majority of woody shrubs produce seeds possessing some form of dormancy at maturity (Baskin & Baskin, 2014). This represents a sig nificant obstacle to overcome in the res toration of arid and semiarid lands. Al though protocols exist to identify seed dormancy syndromes and appropriate methodologies for dormancy alleviation (Kildisheva et al., 2020; Pedrini & Dixon, 2020), in most dryland regions reliable information about seed dormancy and its effective alleviation remains entirely lacking.
In the Monte Austral, of southern Ar gentina, an arid and highly degraded region where ERR is notably hampered by scarce information regarding seed dormancy syndromes, we suggest that a “professional intelligent tinkering” approach (Murcia & Aronson, 2014) is needed. This approach suggests “shortcuts” may be taken in ERR deci sion-making where knowledge is incom plete but inferences can be made from preliminary information. For example, the selection of germination treatments inferred to be likely successful from pre vious studies in climatically-similar re gions.
Even where seed dormancy can be reliably and effectively alleviated, a sec ond practical problem that restoration practitioners face is that successful ger mination procedures alone may be insuf ficient as a means for selecting the best species for ERR in severely-degraded drylands. For example, species produc ing seeds that can be reliably germi nated in the nursery may possess traits resulting in poor seedling emergence and establishment (Long et al., 2015), or unsatisfactory rates of growth and devel opment when outplanted as seedlings in drylands (Pérez et al., 2019a; Pérez et al., 2020).
To explore how to better integrate seed germination characteristics with other important restoration-relevant metrics, such as successful incorporation in out planting activities, our objectives in this study were to: a) experimentally evaluate seed germination success in a group of native species from Monte Austral we consider DFS candidate using a suite of well-known, simple, pre-germination treatments, and b) assess the suitability of selected species for ERR in the light of results of germination trials togeth er with available information on field performance (survival and plant cover in outplanting or direct seeding) in our study region
Material and Method
Study area
Geography and land use
The arid Monte biome covers ca. 466,975 km2 along a 2400 km NW-SE diagonal, extending from 24°S in a near ly subtropical region to 44°S in a rela tively cold region (Morello et al., 2012). The study area is located in the Neuquén Province (Argentina) (Figure 1), which harbors the southern, extra-tropical subdivision of the Monte known as the Monte Austral.
Here, approximately 32% of the territory suffers from seri ous ecological and economic degrada tion, mainly caused by extensive, poor ly-managed cattle ranching and poorly or unregulated industrial drilling for gas and oil (Mazzoni & Vazquez, 2009; Mo rello, 2012).
In Neuquén Province, recent, ram pant oil and gas sector activity has led to partial removal and degradation of vegetation over thousands of hectares, highly negatively impacting biodiversi ty and the health of ecosystems and hu man communities (Mazzoni & Vazquez, 2009).
Climate and vegetation of the Monte Austral
The mean annual temperature in Monte Austral is 15 °C with high seasonal varia tion (Morello et al., 2012). Data recorded at our seed collection sites over four years have shown a mean temperature during summer months of c. 25 °C, and c. 9 °C in winter. Over the last 20 years, mean annual rainfall in the study area was 152 ± 60.3 mm, ranging from 52 to 250 mm annually (data from “La Higuera” meteo rological station, AIC, personal communication July, 2020).
The dominant vegetation type in Mon te Austral is steppic, with shrub-domi nated patches alternating with patches of very sparse plant cover (Busso & Bonvis suto, 2009). The dominant shrub species include Larrea divaricata Cav., L. cunei folia Cav., and L. nitida Cav. (Zygophyllaceae), Monttea aphylla (Miers) Benth. & Hook. var. aphylla (Scrophulariaceae), Atriplex lampa (Moq.) D. Dietr. (Chenopodiaceae), Lycium chilense Miers ex Bertero (Solanaceae), and Prosopis flexuosa DC. var. depressa Roig (Fabace ae). Among low-growing perennial life forms, three clump grasses are common, namely Pappostipa speciosa var. speciosa (Trin. & Rupr.) Romasch, Panicum urvi lleanum Kunth., and Poa ligularis Nees ex Steud, as well as the subshrub Hyalis argentea D. Don ex Hook. &Arn. (Aster aceae) (Busso & Bonvissuto, 2009).
Species richness, seed collection and storage
To describe the richness of shrub spe cies, nine 15-meter-long intersecting transects were randomly laid out in seed collection sites (Figure 1).
Seed collection was carried out from November 2010 to January 2013 fol lowing standard protocols for ecological restoration work (i.e., harvesting seeds from at least 30 to 50 plants in a popula tion without surpassing 20% of the avail able seeds per plant; Pedrini & Dixon 2020). Following collection, seeds were air dried at room temperature in a venti lated space, manually cleaned to remove impurities, and then stored at -18°C until germination tests were conducted (between May and August 2013).
Seed dormancy alleviation treatments and experimental design
Prior to carrying out germination tests, seed fill was determined manually by squeezing seeds between fingers or fin gernails. All unfilled seeds were discard ed. Rates of germination were then eval uated quantitatively using a completely randomized design with four pre-ger mination treatments and a control group (C), each with three replicates of 30 ran domly-selected seeds.
The selected treatments were chosen on the basis of the hypothesis of pre dominance of physiological (PD) and physical dormancy (PY) in cold- or tran sitional, warm-cold-deserts such as the Monte Austral. Seed were exposed to ei ther a cold-wet treatment for 7 or 30 days (CW7 and CW30), or chemical scarification for 5 or 45 minutes (CS5 and CS45); the treatment times considered effective for most species of the region tested to date (Paredes et al., 2018).
For CW7 and CW30 treatments, seeds were distributed in a single layer on a polystyrene tray lined with thin paper and a layer of paper napkins moistened with water. These were covered with a second layer of paper, wet cotton, and another tray as a lid. The trays were re frigerated at 4°C for 7 or 30 days. For chemical scarification, the seeds were immersed in sulfuric acid (H2SO4; 95- 98% purity, Cicarelli laboratory) for pe riods of 5 or 45 minutes.
After the treatments, seeds were placed in Petri dishes lined with a moistened filter paper and placed in a germination chamber (Mechatronics Services brand) under controlled conditions. We used alternating temperatures between 10 ± 1°C for 12 h in darkness and 20 ± 1°C for 12 h, corresponding to the light pe riod. This temperature and light regime emulates climatic conditions in the study region during fall, when seed germination likely occurs naturally (Páez et al., 2005). The criterion for germination was emergence of the radicle. Germination was monitored every two days for 42 days, until no further germination was recorded.
Statistical analysis
To identify groups of species with a sim ilar germination response to dormancy alleviation treatments, we performed a correspondence analysis on species with over 50% germination rates using InfoS tat software (Di Rienzo et al., 2014). A binary logistic regression (SPSS Statistics 25, IBM) was used to assess the main interaction effects of CW and CS treat ments on the successful outcome of seed germination. When no significant differences in germination between treat ments were found, a mean germination time (MGT) was used as supplementa ry information to identify the average number of days it took for a single seed to germinate according to the following formula:
Where fi is the number of days elapsed since the start of the germination test and xi is the number of seeds that germi nated within consecutive time intervals.
MGTs were analyzed using one-way ANOVA with Tukey post-hoc tests using the InfoStat software, with a uniform sig nificance level set at 0.05.
Results
A total of 27 shrub species, belonging to 9 botanical families and 23 genera, were recorded in the study area. However, necessary amounts of seeds for labora tory testing could be collected for 16 of these species (Table 1).
We also consid ered results for two native species - one small tree, Parkinsonia praecox and one shrub, Senna aphylla - collected in the same area, for which germination suc cess have been previously reported using the same experimental procedure and design as we describe here (see Paredes et al., 2018).
Germination response to seed dormancy alleviation treatments
A first response group (G0) was identi fied for the five species that showed al most no germination even after 42 days. Of these, Larrea nitida and Monttea aphylla did not germinate at all, while L. cuneifolia, L. divaricata, and Neospar ton aphyllum all had germination proba bility <0.1 (Table 2).
Correspondence analysis made it pos sible to divide the remaining 11 species into three additional subgroups, named G1, G2, and G3, all showing germination success over 50% with at least one exper imental treatment (Figure 2).
The differ ent probabilities of germination success for each species according to treatment in G0, G1, G2 and G3 are shown in Fig ure 3.
Subgroup G1
The four species in this subgroup showed uniformly medium and high germination rates, and showed no sig nificant differences between treatments. In the case of B. spinosa, the lowest ger mination probability (GP) was observed for treatment CS45 (0.19, 0.11-0.27, Ex p(B) = 2.471 [1.357, 4.500], Wald = 0, P < 0.005). All other treatments yielded high GP and did not differ statistically significantly from the control (Table 2).
Most seeds of G1 species germinated within the 30-day stratification period in CW30, and thus CW30 was not included in MGT calculation. Among other treat ments, lowest MGT was reported for all G1 species in CW7 (Figure 4).
Subgroup G2
Treatments CS5 and CS45 yielded sig nificantly greater GP compared with controls for P. flexuosa var. depressa (hereafter P. flexuosa), P. praecox and S. aphylla. Germination probability for untreated P. flexuosa and P. praecox seeds was relatively low (0.30, 0.21-0.39, and 0.28, 0.19-0.37, respectively) but was three-fold higher in CS5 and CS45 for P. flexuosa (GP = 0.97, 0.93-1.00, Exp(B) = 219.077 [28.976, 1656.38], Wald = 27.265, P < 0.005) and nearly three-fold higher in CS45 for P. praecox (0.87, 0.81- 0.94, Exp(B) = 22.000 [9.821, 49.281], Wald = 56.430, P < 0.005). GP was very low for untreated G. chiloensis seeds (0.03, 0.00-0.07), but improved mark edly in CS45 (0.51, 0.41-0.61, Exp(B) = 93.045 [12.420, 697.080], Wald = 19.465, P = 0.002). GP was also higher in CS45 for A. undulata (0.60 0.50-0.70, Exp(B) = 2.471 [1.357, 4.500], Wald = 8.743, P = 0.003) than in controls (0.39, 0.28-0.48), but decreased significantly in CW30 (0.07, 0.02-0.13, Exp(B) = 0.097 [0.0036, 0.263], Wald = 21.043, P < 0.005).
Subgroup G3
GP for untreated S. filaginoides seeds was very low (0.06, 0.01-0.11), but increased almost ten-fold in CS5 (0.52, 0.42-0.62, Exp(B) = 23.500 [7.945, 69.511], Wald = 32.554, P < 0.005). Untreated S. sub ulatus seeds germinated relatively well (0.54, 0.044-0.64) but GP was signifi cantly improved in CS5 (0.77, 0.68-0.85, Exp(B) = 2.929 [1.533, 5.595], Wald = 10.584, P = 0.001). GP in S. subulatus was markedly decreased in CW7 (0.39, 0.29-0.49, Exp(B) = 0.532 [0.294, 0.963], Wald = 4.338, P = 0.037) and CW30 (0.34 0.24-0.44), Exp(B) = 0.418 [0.229, 0.765], Wald = 8.010, P = 0.005). After 45-minute chemical scarification (CS45) all seeds of all G3 species exhibited em bryo damage and, consequently, no ger mination was recorded (Table 2).
Species performances in ERR
Although field survival and growth data were obtained from the literature, most of them were limited to particular soil properties (see Discussion). We found precise information from field studies on survival for 9 of 16 species, all of them with high survival rates (>75%) except for S. subulatus whose best performance was associated with non-alkaline soils. As regards to growth or plant cover only five species have been evaluated, four attained relatively high plant cover com pared to other species planted simulta neously (Table 3).
Discussion
With respect to seed collection, great variability in fruit production has been observed in tree and shrub species native to arid and semiarid lands. In the Monte region, Dalmasso & Anconetani (1993) reported seed production of P. flexuosa varying from 80,000-800,000 seeds·ha-1 in different years. Resource limitation, the time of rainfall, frost and wind oc currence, and soil water content have all been suggested as proximate causes of differing seed abortion rates (Villagra, 2000). Possibly, some of these reasons are the causes that determined that we could not gather enough seeds to carry out more detailed experiments for 11 of the 27 candidate species.
The “G0” subgroup
Five out of the 16 species tested showed negligible response to the dormancy alle viation treatments applied. Recent studies have indicated that for Larrea species, mechanical scarification can be effective (Fernandez et al., 2019), and a novel low-cost scarification technique with a drill and sandpaper shows promise for appli cation for large volumes of seed such as will be needed for large-scale ERR (Her nandez et al., 2020). There are no previ ously reported studies on germination for N. aphyllum or M. aphylla. In relation to outplanting performance, only L. di varicata has been studied, with reports of very strong survival results of ca. 79% (Pérez et al., 2019b). Thus, this is the only species in this group that we can safely called suitable and promising for large-scale ERR restorative practices.
The “G1” subgroup
In this subgroup, medium germination rates were obtained (ca. 50%) for A. lam pa in all treatments applied. Previous researchers have reported higher ger mination rates for this species (ca. 80%) after manually removing the bracteoles (Bonvissuto & Busso, 2007). However, this procedure would be prohibitively expensive for the large volume of seeds required in large-scale ERR. The other three species in this subgroup (B. spino sa, E. ochreata, and H. argentea) showed very high germination rates (>95%) without treatments of any kind. High rates of germination from seeds in this manner may favor natural regeneration where reproductively-mature individu als of these species could be established to provide a source of seeds in the restor ing site.
All the species in this subgroup ger minated within 30 days of the cold-wet treatment, with the shortest mean germination time (MGT) in the CW7 treatment (Figure 4). This supports a seed recruitment niche in these species matching the thermal and moisture con ditions observed during the autumn pe riod where natural seed germination has been observed. With respect to species performance in the field, A. lampa and H. argentea showed very high survival rates (84-91%), and relatively good plant cover compared to other species plant ed at the same time (Pérez et al., 2019a; 2020). Outplanting of B. spinosa in com pacted soils yielded high survival rates (85%; Pérez et al., 2020), although this species generally grows quite vertically with few ramifications and bearing only very small leaves, and thus covers very little ground even when mature (Pérez et al., 2020).
The “G2” subgroup
In this group all the species, even from different families, such as Chenopodia ceae, Fabaceae, and Asteraceae, showed germinations rates over 50% with a single treatment (CS). Previous studies reported higher germination for A. un dulata following manual removal of the bracteoles (Piovan et al., 2014). Howev er, this labour-intensive approach is like ly impractical for large-scale ERR. Acid scarification of seeds in the manner we tested yielded acceptable seed germina tion results, and partially resolves issue of scale in application to large quantities of seeds.
Regarding species suitability for ERR, G. chiloensis and P. flexuosa both ex hibited high survival rates and relative ly high growth rates compared to other species outplanted concurrently (Becker et al., 2013; Pérez et al., 2019b). How ever, P. praecox and S. aphylla had no tably higher survival rates after being outplanted, although P. praecox showed relatively little plant cover in very com pacted soils after five years (Pérez et al., 2020). No information could be found on field survival or plant cover for A. un dulata and S. aphylla, and we suggest further study is required for these taxa.
The “G3” subgroup
CS5 yielded the highest germination rate for two congeneric species in this group (50% and 77% for Senecio filagi noides and S. subulatus, respectively), and this treatment should be preferred over the standard procedure of nicking seeds (i.e. cutting the testa with a scalpel; Kildesheva et al., 2020), which achieves similar results but it is very time-con suming (Masini et al., 2016). Senecio subulatus plantations have been recom mended for non-alkaline soils (Pérez et al., 2019a) and new studies are required to assess survival and growth attributes for S. filaginoides in order to assess their suitability for large-scale plantations and holistic restorative programs.
Final remarks
In many arid and semiarid regions, es pecially those in developing countries, embarking on the long process of ERR can be extremely complex and costly, especially in the face of extreme and unrelenting desertification processes. Fortunately, more and more easily ac cessible protocols and recommendations are being developed to address and solve technical and strategic difficulties faced by ERR practitioners. Furthermore, the positive benefit-cost ratios of such in vestments, and the worldwide trend to step up national commitments to ERR -are changing the picture. But, for now, the problems of getting restoration pro cesses in gear remain daunting. In this case study, we employed a tool that we consider very useful for dryland resto ration practitioners, namely professional intelligent tinkering (Murcia & Aronson, 2014). Simple trial and error experi ments on a small scale are the hallmark of this approach. When good results are obtained, then more detailed scientific experiments can be undertaken as a fol low up in view to solid validation.
In this manner ERR activities can progress concurrently with on-site re search undertaken to fill knowledge gaps in best practice, and/or with cur rent practices modified and improved through input of knowledge and know-how from external sources. We applied this conceptual framework to evaluate germination, using hypotheses about seed dormancy type and dormancy alle viation methodologies for groups of spe cies rather than empirically evaluating each and every candidate species. This approach allowed us to achieve accept able germination success (50-100%) in 11 out of 16 studied species using only two simple dormancy alleviation treat ments. These results, coupled with field performance data and recently-pub lished information on seed germination for three additional species, allowed us to propose eight of the 16 studied spe cies as DFS candidates for our region based on their seed germination traits, high survival rates when outplanted, and potential to rapidly achieve high rates of plant cover.
In conclusion, our study offers one example of how some typical knowl edge gaps constraining ERR can be side-stepped to allow restorative activities to progress even when complete informa tion on best-practice approaches is lacking.