Introduction

Currently, several major global challenges are related to the environment and the way land is used and managed. Among the most urgent global challenges are the need to mitigate climate change, battle population growth, desertification and land degradation (^{Mirzabaev et al., 2019}; ^{Smith et al., 2020}). Processes at scale risk the provision of ecosystem services (^{Millennium Ecosystem Assessment, 2005}; ^{The Lancet Planetary Health, 2018}). Moreover, unsustainable practices of natural resource use contribute to the loss of biodiversity and natural habitats (^{Rechkemmer et al., 2016}), in addition to the impact of the COVID-19 pandemic, whose future implications are still unknown (^{Forster et al., 2020}).

The Sonoran Desert is not exempt from current environmental problems due to its arid and semi-arid characteristics (^{Castellanos-Villegas et al., 2010}), such as variable precipitation, high temperatures in summer and little food for herbivores in the dry season (^{Carevic et al., 2017}). Regeneration is more difficult to take place in this region, especially in parts of the desert with degraded soils (^{Whitford, 2002}). Biodiversity risks, recovery time and low resilience in this area affect, specifically, endemic plant species (^{Van Devender et al., 2010}).

Ironwood (Olneya tesota A. Gray) is an endemic plant of the Sonoran Desert, which has strong positive effects on soil fertility and hydrology (^{Celaya-Michel et al., 2015}; ^{Castellanos et al., 2016}). This tree is considered of great ecological importance because it facilitates the establishment of other plants under its canopy (^{Suzán et al., 1996}). However, it is a plant that does not produce seeds every year (^{Shreve & Wiggins, 1964}), but once established it can last for several hundred years (^{Turner et al., 1995}).

Currently, the ironwood in Mexico is under the protection status of NOM 059 (^{Secretaría de Medio Ambiente y Recursos Naturales [SEMARNAT], 2010}; ^{Van Devender et al., 2010}) due to the decline of its populations due to deforestation, over-harvesting and little natural regeneration in ecosystems with progressive degradation (^{Castellanos-Villegas et al., 2010}). For hundreds of years, ironwood has been used by ethnic groups and settlers as a source of food, wood, shade, forage, and other uses (^{Phillips & Comus, 2000}).

In Mexico, government programs have been developed to support reforestation of important areas (^{Dorado & Arias, 2006}). This is done by planting seeds in areas of water accumulation, transplanting nursery-grown plants and using netting to protect against herbivory (^{Comisión Nacional Forestal [CONAFOR], 2010}). However, reforestation results have not always been favorable due to a confusing set of interactions and feedbacks (^{del Campo et al., 2021}). Furthermore, initiatives to restore natural ecosystems have had little success in arid and hyper-arid areas, as the establishment of natural seedlings is affected by drought patterns and climate extremes (^{Carevic et al., 2017}), highlighting water availability, elevated temperatures, herbivory (^{Archer & Pyke, 1991}; ^{Bowers et al., 2004}; ^{Moreno et al., 2017}), and typical strategies used in restoration such as site selection, objectives, and costs (^{Li & Gornish, 2020}) as limiting factors.

So far, no work has been found on direct sowing of ironwood seeds in degraded lands, which would provide more effective restoration strategies in lands used for cattle. For decades, it has been known that Sonora has the potential to develop commercial forest plantations using native plants to restore degraded agricultural and cattle lands (^{CONAFOR, 2010}). However, little information currently exists on the establishment of native plants, both shrubs and forage trees in the Sonoran Desert (^{Martín-Rivera et al., 2001}).

Therefore, the objective of the present study was to evaluate the effect of ironwood seeds sown under different land use conditions regarding germination, survival and growth in Hermosillo (Sonora, Mexico).

Materials and methods

The study area is located in the Departamento de Agricultura y Ganadería (DAG) of the University of Sonora, Hermosillo, Sonora, Mexico, with an area of 244 ha (29° 01’ 13’’ LN and 111° 08’ 05’’ LW, at 191 m a. s. l.). The climate of the site is very dry and warm, with average annual temperatures of 23.1 °C (with maximum temperatures in June of 49 °C and minimum temperatures in February of -3.5 °C) (^{Comisión Nacional del Agua [CONAGUA], 2014}). Precipitation occurs mainly during the summer monsoon, between July and August (^{Instituto Nacional de Estadística y Geografía [INEGI], 2014}), with average annual precipitation of 378 mm. Precipitation information for recent years was collected from the nearest station of the weather station Red de Estaciones Meteorológicas Automáticas de Sonora (Table 1). Soil type is predominantly sandy loam (^{Food and Agriculture Organization of The United Nations [FAO], 2015}).

Seeds of ironwood (Olneya tesota) were collected at the study site mid 2016, with a collection permit from SEMARNAT because this species is included in the list of NOM-059-SEMARNAT-2010.

Fifty ironwood seeds were sown per patch of soil with different characteristics (degraded, conserved, agricultural and nursery), in an area of 1 m² per patch. Each seed was considered a replicate. The soil was manually loosened with pickaxe and seeds were sown at a depth of 6 mm, which were covered with soil from the site (^{Bonner & Karrfalt, 2008}). Sowing was performed at the beginning of the summer rainy season of 2017, 2018 and 2019. The variables evaluated were germination (seedling emergence), survival (at 90 and 365 days) and growth (maximum height at 365 days measured with a flexometer).

Grazing areas of the DAG were chosen, where the predominant vegetation was arbosufrutescent shrubland, but for several decades buffelgrass (Pennisetum ciliare [L.] Link) has been planted for cattle grazing. Currently, judging by the large amount of bare soil (70 %, determined using Canfield’s [1941] methodology), there is degradation, erosion and compaction. The site is surrounded by barbed wire supported by metal posts, and there is cattle grazing when rainfall favors plant growth.

The “degraded” plot was contrasted with a “conserved” plot, which was a 20-year-old botanical garden (^{Celaya-Michel et al., 2019}) located on 2 ha excluded from cattle grazing, where bare soil was 10 % (determined with the same methodology) and there was higher density of trees, shrubs, herbs and grasses compared to the degraded plot.

In both ecosystems, degraded and conserved, small herbivores were observed such as hares (Lepus californicus), jews (Xerospermophilus tereticaudus), rabbits (Sylvilagus audubonii), mule deer tracks (Odocoileus hemionus), lizards (Callisaurus draconoides) and ants (Pogonomyrmex rugosus).

The DAG nursery was used for seed germination under controlled conditions of humidity and without risk of herbivory. Fifty seeds were sown in an equal number of plastic pots containing 1 kg of local soil and were placed under a shade netting. Two weekly irrigations of 250 mL were applied to each pot. In addition, ironwood seeds were sown in deep agricultural soil with drip irrigation within a 3.35 ha triangular plot protected with 2 m high wire netting formed by 5.7 cm squares to limit the entry of domestic animals and wildlife. The irrigation rate applied was 20 mm per week, and weekly irrigation was suspended when weekly precipitation accumulated 20 mm or more.

Four experiments were established at different times (July 2017, June 2018, August 2018 and July 2019), and the performance of ironwood seeds was contrasted in degraded, conserved, agricultural and nursery ecosystems. In the degraded and conserved ecosystems, patches of bare soil, grasses (native and exotic), shade under tree and water accumulation areas were used. The bare soil was covered with dead plant material, consisting of branches from local trees, to find more favorable conditions for seed development, which was called bare soil plus branches. The grass patches were divided into native, where Rothrock’ grama (Bouteloua rothrockii Vasey), was predominant, and exotic, where buffelgrass (Pennisetum ciliare (L.) Link) was present. Daily observations were made during the 10 days following each precipitation event greater than 5 mm, and only on Mondays and Fridays in the weeks when there was no precipitation.

In the penultimate experiment, the patches established were covered with plastic netting made of 1 cm squares, and in the last experiment, 0.5 cm metal netting was used only on bare soil. In the latter, 10 mm of weekly irrigation was added, which was suspended when there were accumulations of 20 mm or more of weekly precipitation. It is important to clarify that the same seed lot was used in all experiments.

The response variables evaluated were germination and growth (plant height). Statistical analyses were performed with the JMP program version 10 (^{Lehman et al., 2013}). With the survival data, a contingency table was performed with the chi-square test (χ²) according to ^{Infante-Gil and Zarate-de Lara (2000)}. The information corresponding to growth (plant height at one year) was contrasted by analysis of variance, prior analysis of normality with the Shapiro-Wilk test, and a Tukey mean comparison was performed. In all cases, a statistical significance lower or equal to 5 % (P ≤ 0.05) was established.

Results and discussion

The results of the experiment initiated in July 2017 show that ironwood seeds germinated more in nursery than those planted in degraded soil patches (68 % germination, χ²= 92.72, P () 0.0001; Table 2). Ironwood plants in the nursery had a survival rate of 60 % at one year, while plants in degraded soil were unable to reach 365 days. This mortality was in a period of less than 90 days, except for one plant in degraded soil under a tree. Plants in nursery had regular watering, which may explain their high survival, while the plants in the degraded soil patches died in the first weeks, due to the low humidity in the first centimeters of the soil and the herbivory observed. Rainfall in July 2017 was extraordinary for that month (164 mm; Table 1); however, from August to December of the same year it was only 77.7 mm followed by dry months from January to May 2018. For degraded soil patches, plant growth was not included because there was no survival (Table 2).

The results of the experiment initiated in June 2018 show that ironwood seeds under irrigated agricultural soil conditions and soil conserved under tree germinated better than those sown in the other soil patches (84 and 74 %, respectively) (Table 3), despite good rainfall from July to October 2018 (Table 1).

Survival at 90 days was zero in many patches studied, including bare soil of the degraded and conserved ecosystem, with grass of degraded and conserved ecosystem, and in accumulation zone of degraded ecosystem. The survival of the bare patch with branches of the conserved ecosystem, the patch under tree of the conserved ecosystem and the agricultural soil with irrigation was 2, 32 and 84 %, respectively. At 365 days the highest survivals were from agricultural soil with irrigation (74 %), soil under tree in conserved ecosystem (30 %) and bare soil with branches in conserved ecosystem (2 %) (Table 3). Regarding growth, few data were available for comparison due to survival in most patches, only plants from agricultural soil with drip irrigation were significantly (P () 0.0001) taller at 365 days than plants grown under tree in a conserved ecosystem (84.7 and 20.2 cm, respectively)

In August 2018, limiting herbivory access with the use of netting in degraded soil patches led to increased germination, exceeding bare soil without netting, accumulation zone (72 %), bare soil (64 %), and soil with native grass (64 %) (χ²= 30.11, P () 0.0001; Table 4). The bare soil patch without netting showed the lowest germination (20 %). Germination, like survival, increased when the net was placed in several patches, especially in the accumulation zone, bare soil, soil under tree and with native grass, reaching one year of survival. However, significant differences (P () 0.05) were found in the average plant height at 365 days (Table 4).

The results of the July 2019 experiment showed that protection with netting improved germination (χ²= 57.43, P () 0.0001; Table 5), compared to unprotected and irrigated treatments. In 2019, the accumulated rainfall was 250 mm from July to September, which is uncommon (Table 1). This may explain the little difference between soil patches with netting, with and without irrigation.

Several investigations highlight the importance of ironwood as a nurse plant, favoring the establishment of other species under its canopy (^{Hutto et al., 1986}; ^{Carrillo-García et al., 1999}; ^{Carrillo-Garcia et al., 2000}; ^{Tewksbury & Lloyd 2001}; ^{Suzán-Azpiri & Sosa, 2006}; ^{Hinojo-Hinojo et al., 2013}). However, no studies were found regarding the germination of ironwood seeds in degraded soils. In general terms, some authors mention that ironwood seeds do not require pre-germination treatment, and that under natural conditions their germination has been observed after rainfall (^{Shreve & Wiggins, 1964}). These seeds begin to germinate 18 to 24 h after rainfall saturates the soil, and seedlings emerge in approximately six days (^{Bonner & Karrfalt, 2008}).

Seed germination rate was 26 % or lower (Tables 2, 3, 4 and 5) for unprotected degraded ecosystem patches evaluated from 2017 to 2019. Something similar occurred in the conserved ecosystem patches, except for the under-tree system, where 74 % germination was recorded (Table 3). In 2018, there was seed production, and it was observed that in the conserved ecosystem there were ants displacing seeds, prior to the start of the experiment. Perhaps the needs of the ants near the tree in the conserved ecosystem were covered by the natural productivity of the ecosystem, which reduced the extraction of seeds sown at this site. Nevertheless, in patches of the degraded ecosystem, ants were observed scavenging and removing swollen seeds, as well as seedlings with herbivory damage and holes in the soil. To the north of the study area, excavation by rodents to remove buried seeds has been reported, and the application of rodenticides to prevent seed loss has even been suggested (^{Archer & Pyke, 1991}). In Asian deserts, excavations made by mammals to remove seeds from the soil have been reported (^{Gutterman, 2012}).

In deserts, seeds represent a valuable resource for rodents, ants and birds (^{Keddy, 1989}; ^{Phillips & Comus, 2000}). About 70 % of seeds in deserts are removed by granivores such as mammals, birds and insects, mainly ants (^{Archer & Pyke, 1991}; ^{Gutterman, 2012}). The main granivores in arid areas of North America are rodents, while in South America, Australia and Africa are ants (^{Archer & Pyke, 1991}).

The species of ants that inhabit the Sonoran Desert, some of which harvest seeds to feed their larvae, preferring seeds of annual species, although they eventually consume perennial plants (^{Phillips & Comus, 2000}). This could be occurring in degraded lands, such as the study area, where ants and rodents consumed part of the ironwood seeds that were sown. Levels of seed predation by rodents are particularly high in disturbed sites (^{Archer & Pyke, 1991}). Determining predation levels in degraded soils by small mammals, ants or other biota may be an interesting aspect for future research.

Seed viability was tested by germination in nursery (68 %; Table 2) and agricultural soil (84 %; Table 3); however, there was no germination in agricultural soil without irrigation, despite rainfall from July to October 2018 (Table 1). This may be due to seed predation by rodents or ants. When protection netting was placed in the patches with degraded ecosystem, germination was between 60 and 78 %, except for the system under buffelgrass, which was 28 % and 48 % under tree (Tables 4 and 5). Therefore, it is believed that there were seed extraction or herbivory of recently germinated seedlings, making it impossible to account for germination.

No published data were found on ironwood growth under direct seeding, only data from nursery-grown plants have been published. ^{Martín et al. (2017)} reported an average plant height of 40 cm in flat soil and 58 cm in water retention furrows two years after transplanting.

At 365 days, survival was observed in three of the four patches studied, and was higher in the bare soil with netting and irrigation (Table 5). In none of the experiments in degraded soil was survival at the first year (Tables 2, 3, 4 and 5). Plants under controlled conditions in nursery and agricultural land with drip irrigation showed 68 and 84 % survival at one year, respectively (Tables 2 and 3). There was 2 % annual survival in the conserved soil when the soil was covered with branches and a microclimate was created, while under the tree canopy survival was 30 % (Table 3). When protection netting was placed on the degraded soil, survival was 20 to 50 %, except for bare soil patches with branches and buffel, both without annual survival (Table 4). Bare soil without netting but irrigated had a survival rate of 8 %, while the treatment with netting and without irrigation had a survival rate of 12 %, and when netting with irrigation was used, survival was 28 % (Table 5).

No publications were found on the survival of ironwood under direct seeding, but ^{Martín-Rivera et al. (2001)}, when evaluating eight transplanted shrub species in Sonora, including ironwood, found an annual survival rate of 55 to 85 %, with 317 mm of rainfall that year. In addition, these authors report that the four-year survival rate for ironwood was 27 %.

In 2001, during the creation of the botanical garden at the DAG, several nursery-developed shrub and tree species, including ironwood, were transplanted. For this, the plants received drip irrigation for one year and survival was 100% (^{Celaya-Michel et al., 2019}). This coincides with the survival rate achieved in agricultural soil with drip irrigation (84 %), although in this case it was direct sowing.

Limited information was found on the survival of ironwood in reforestation. However, ^{Sosa-Castañeda et al. (2019)} observed a survival of 34 to 56 % in jelly bean tree (Parkinsonia microphylla Torr.) transplanted with different protections against herbivory and 1.8 % in plants transplanted without protection. In Arizona, USA, ^{Bowers et al. (2004)} found 3.7 % survival per year in 15 perennial species from the Sonoran Desert, germinated directly in soil and irrigated with local rainfall. The causes of mortality in this area were herbivory and desiccation.

Ironwood plants that germinated under buffelgrass had a survival of zero at 90 days (Table 4), which could be due to competition with buffelgrass for available moisture, as reported by ^{Eilts and Huxman (2013)} in an adult tree legume, which showed branch sacrifice when individuals of buffelgrass grew under its canopy. The survival rate of the plants surviving the first year (Table 3) was 80 to 100 % at the end of the study (i.e., when they were two and a half years old), while in 2018 the survival rate was 0 to 52 %, being the plants in bare soil with netting where the value was 0. The growth of surviving plants at 365 days was significantly (P () 0.0001) higher for ironwood plants grown in degraded soil protected with netting and irrigated (30.3 cm vs. 12.2 cm in plants irrigated and without netting, and 10.9 cm for unprotected and irrigated plants) (Table 5).

The growth of the plants that survived the year ranged from 12.7 to 30.3 cm in the different patches studied, and the growth under agricultural conditions with drip irrigation was 84.7 cm (Tables 2, 3, 4 and 5).

The Mexican ironwood has been described as a slow-growing plant with a longevity of hundreds of years once it becomes an adult plant (^{Nabhan & Carr, 1994}). However, no records were found on the annual growth rates of ironwood from plants germinated directly on soil. In a study in north-central Sonora, they report nursery growth of eight shrub species for one year, including ironwood, from 25 to 75 cm, and subsequent growth in 4 years from 30 to 68 cm (^{Martín-Rivera et al., 2001}). In the botanical garden of the DAG, ironwood established with drip irrigation during its first post-transplanting year had an average growth of 39 cm per year during the 14 subsequent years, where irrigations were seasonal rains (^{Celaya-Michel et al., 2019}).

Natural revegetation of degraded land on the coast of Hermosillo is very difficult, as it requires a combination of factors, such as seed production (^{Shreve & Wiggins, 1964}), above average rainfall for several years (^{Bonner & Karrfalt, 2008}), soil patches with suitable conditions, and overcoming herbivory (^{Celaya-Michel et al., 2019}). Artificial revegetation through direct seeding can be supported with soil decompaction, seed sowing (^{CONAFOR, 2010}), seed protection on soil, microclimate generation and soil maintenance to have water accumulation zones (^{Martín-Rivera et al., 2001}). Commercial forest plantations in abandoned agricultural fields could be another alternative for the inclusion of native plants, such as ironwood, with seasonal drip irrigation support to increase their annual growth rate (^{CONAFOR, 2010}).

Conclusions

Ironwood seed germination is mainly affected by herbivory, which is more frequent in degraded soils than in conserved ones. The survival of ironwood seedlings depends on herbivory activity, soil water availability and protection provided by conserved ecosystems. Netting increases germination, survival and height of ironwood seedlings in degraded soils. Growth in the first year in arid lands is significantly below the potential growth in agricultural soils and with drip irrigation. It is possible to initiate ironwood forest plantations in agricultural soil with drip irrigation.