Response of seed germination and seedling emergence of Haloxylon ammodendron to rain frequency and temperature change from four desert ecosystems, Northwest China

Abstract Climate change will result in variation of rain frequency and amount and warming in arid zones, which is expected to affect seed germination and seedling emergence in desert ecosystems. However, the effects of unpredictable rainfall and increasing temperature on seed germination and seedling emergence of dominant desert plants remain unclear across different deserts, which are important for population regeneration and community succession in desert ecosystems. Seed germination and seedling emergence of Haloxylon ammodendron across four deserts in Northwest China were examined at different rain frequencies with same total amount, and constant and alternating temperatures, to investigate their response to climate change. Rain frequency determined seed germination and seedling emergence of H. ammodendron in the Tengger Desert, Badain Jaran Desert, Gurbantonggut Desert and Mutthar Desert, which was maximal at rain frequency of 10 times per month and decreased with a decrease of rain frequency. Temperature was not a restricting factor for seed germination of H. ammodendron in the Tengger Desert, Badain Jaran Desert and Gurbantonggut Desert, varying from 10 °C to 25 °C and from 20/10 °C to 30/15 °C, respectively. However, the highest temperature of 25 °C and 30/15 °C inhibited seed germination of H. ammodendron in the Mutthar Desert. Thus, H. ammodendron has an opportunistic germination strategy. Under climate change in the future, seed germination and seedling emergence of H. ammodendron would be restricted by the combination of less frequent rainfall and increased temperature in desert ecosystems. The regeneration of the H. ammodendron community should be promoted by irrigation and seedling transplant.


Introduction
Global climate change due to the greenhouse effect of elevated CO 2 includes changing precipitation regimes and warming (IPCC 2013). A desert ecosystem is highly sensitive to climate change because plant species are already struggling in an extreme environment, including drought and high temperature. Since water and temperature are two key factors for plant survival and growth in a desert ecosystem, the changes of hydrothermal availability are expected to have an influence on the structure and function of the desert plant community (Liu et al. 2016;Bai et al. 2020;Wang et al. 2018a).
Seed germination is the key stage in the establishment of plant species, which is also the most sensitive stage to environmental change in the life cycle . Plant species have different seed germination strategies to ensure their survival in the desert ecosystem (Gutterman 2002). Some desert plants have an opportunistic strategy and produce seeds that germinate soon after rain in the growing season (Tobe et al. 2005b). However, other desert plants, especially perennial herbs and shrubs, have a cautious strategy (Commander et al. 2017) and produce few seeds that germinate only in specific conditions. Seed germination of desert plants often is restricted by many environmental factors, such as moisture, temperature, light and sand burial (Li et al. 2012;Zhu et al. 2014), depending on the species. Germination time may be limited to spring or autumn, or seeds may germinate throughout the growing season . Therefore, investigating seed germination of desert plants will increase our understanding of their adaptive strategy to arid climate and result in sustainable management practices of desert vegetation.
In desert ecosystems, precipitation is usually low and unpredictable in its spatiotemporal pattern (Gutterman 2002), and mainly occurs as sporadic rainfall during the growing season, restricting seed germination and seedling emergence (Gillespie et al. 2004;Tobe et al. 2006). It has been predicted that under global warming, less frequent, but higher precipitation events will increase in the future (Allan and Soden 2008;IPCC 2013). The frequency, intensity and spatial pattern of rain have important effects on seed germination and seedling establishment in the desert (Gutterman 2002). Seeds of desert shrubs may need continuous rain events or heavy rainfall to trigger seed germination (Schwinning and Sala 2004). For example, high seedling emergence of two Artemisia semi-shrubs, Caragana korshinskii and Hedysarum fruticosum occurred at 10 mm, 5 and 7.5 mm and 5 mm rain every 3 days in Mu Us Sandy Land (Zheng et al. 2004(Zheng et al. , 2005. Seedling emergence of three Artemisia semi-shrubs was maximal with an initial 16 mm and subsequent 3 mm of irrigation at 1-day intervals in deserts of Northwest China (Tobe et al. 2006). Watering level for maximal seedling emergence of C. korshinskii, Hedysarum leave and Artemisia ordosica in Mu Us Sandy Land was 10 mm, 10-20 mm and 15-20 mm, respectively (Zheng et al. 2009). Moreover, with increased rain amount, seed germination and seedling emergence generally increased, such as three Artemisia semi-shrubs (Tobe et al. 2006) and Artemisia sphaerocephala in Mu Us Sandy Land (Yang et al. 2012), and Reaumuria soongarica in Badain Jaran Desert (Shan et al. 2018). Rain frequency also influenced seedling emergence. With a decrease in rain frequency, seedling emergence decreased, e.g. A. ordosica (Tobe et al. 2006). However, most studies about seed germination and seedling emergence in different rain patterns were performed for one species or for interspecific variation in one area. For widely distributed species across different areas, we still do not know the intraspecific variation of seed germination under different rainfall regime.
Temperature is another environmental factor that determines seed germination (Dürr et al. 2015). Moreover, the seasonal change of temperature primarily regulates germination in the field. Seeds of most shrubs in temperate desert ecosystems generally germinate over a wide range of temperatures, varying from 3 °C to 30/15 °C . Seeds of some desert shrubs germinated well when they were incubated at constant temperature. For example, most seeds of two Haloxylon shrubs in China germinated at 5 °C to 30 °C (Tobe et al. 2000). Maximal seed germination of Nitraria tangutorum and Nitraria spaerocarpa occurred at 15 °C to 25 °C and at 20 °C and 25 °C, respectively (Wang and Zhang 2009). In Horqin Sandy Land, seeds of two Artemisia semishrubs germinated at 10 °C to 34 °C (Li et al. 2012); however, seeds of Caragana micophylla germinated well at 15 °C to 30 °C (Lai et al. 2019). Seeds of other desert shrubs germinated well at alternating temperatures. For example, seeds of C. korshinskii from Ordos Plateau germinated at 5/15 °C to 25/35 °C (Lai et al. 2015). Seeds of Artemisia halodendron, C. korshinskii and C. microphylla from the Horqin Sandy Land germinated well at 20/10 °C to 30/20 °C (Lai et al. 2016). Seed germination of 11 species from a desert steppe in Inner Mongolia increased as temperature was increased from 0/12 °C to 15/27 °C, whereas another nine species germinated well at the lowest temperature of 0/12 °C (Yi et al. 2019). Moreover, alternating temperatures were more favourable for germination than constant temperature of some psammophytes, such as C. korshinskii . However, most previous studies about seed germination at different temperatures were carried out for one species or for the interspecific variation in one area. For widely distributed species, there were few studies about the intraspecific variation of seed germination from different areas, e.g. the differentiation in seedling emergence of Coleogyne ramosissima in Mojave Desert and Colorado Plateau (Meyer and Pendleton 2005).
Haloxylon ammodendron is a dominant shrub or small tree in the deserts of Northwest China and central Asia, which is widely distributed in sandy desert, Gobi, clay desert and salt desert (Ma 1991). This species plays an important role in the ecological function of the desert ecosystem, such as sand dune stabilization, carbon sequestration and biodiversity conservation (Gao et al. 2010;Zhu and Jia 2011). Moreover, it has important economic value as the host plant for Cistanche deserticola, a Chinese traditional herb (Ma 1991). The species has been widely used in land desertification control in Northwest China and central Asia. Regeneration of the H. ammodendron community relies on seed germination and seedling establishment. Previous studies showed that seed germination of H. ammodendron was affected by temperature, which was suitable at 10 °C and lowest at 30 °C; however, it was not affected by light or darkness (Huang et al. 2003). High seedling emergence of H. ammodenron resulted from 8 to 20 mm of irrigation (Tobe et al. 2005a). However, we still do not know the effect of rain frequency and alternating temperatures on seed germination of this dominant shrub in the desert ecosystem. To obtain a better understanding of the adaptive strategy of dominant desert shrubs to a harsh environment, we investigated the effects of rain frequency and temperature change on seed germination of H. ammodendron and discussed the possible impact of climate change on its population regeneration. Our research will give theoretical support to the sustainable management of desert vegetation under climate change.

Seed collection
In late October and early November of 2019, seeds (utricles) of H. ammodendron were collected in four desert ecosystems from east to west in China, e.g. Tengger Desert, Badain Jaran Desert, Gurbantonggut Desert and Mutthar Desert, which are the main distribution areas for H. ammodendron in the arid zone of Northwest China. The location and climate of the four seed sources are shown in Table 1. The four deserts have a typical temperate continental climate. Tengger Desert and Badain Jaran Desert are located on Alxa Plateau of Mongolian Plateau, which has less mean annual precipitation than the other two deserts. Gurbantonggut Desert and Mutthar Desert are located in Junggar Basin of central Asia. The mean daily air temperature is highest in July in the Mutthar Desert. Seeds were manually shaken from shoots of H. ammodendron, and then stored in a cotton bag at room temperatures varying from 13 °C to 22 °C for 5 months.
Before this experiment, the pericarp and wing of H. ammodendron seeds were removed manually. Seed mass was measured by an electronic balance (0.01 g) for 1000 seeds with four replicates. The mean mass of 1000 H. ammodendron seeds from the Tengger Desert, Badain Jaran Desert, Gurbantonggut Desert and Mutthar Desert was 3.25, 2.65, 3.63 and 2.73 g, respectively (Table 1).

Experimental design and measurement
Based on the field observation, most seeds of H. ammodendron germinated after rain from April to June. Thus, climate data of four seed sources were analysed to determine experiment conditions. On average, there were 2.00 to 3.71 times rainfall events in the four seed sources in April, 1.71 to 5.00 times in May, and 4.00 to 5.00 times in June (Table 2). Thus, four rain frequencies (10, 6, 3 or 2 times per month) were applied in our experiment, considering historical rainfall data and the decrease of rain frequency resulting from climate change in the future.
Seed germination tests were conducted from 12 April to 13 May in 2020 in a non-heated greenhouse at the Chinese Academy of Forestry. The substrate for seed germination was river sand passed through a sieve with the diameter of 2 mm. Plastic pots (15.6 cm diameter × 13.2 cm height) were filled with sand to within 1 cm of the top, and 25 seeds were planted uniformly with forceps at 1 cm depth in each pot because the highest seed germination percentage of H. ammodendron was obtained at 1 cm depth . There were four pots (replicates) for each treatment, totalling 64 pots in this experiment (4 rain frequencies × 4 seed sources × 4 replicates). A seedling was considered to be emerged when its first foliage leaf was 5 mm above sand surface. Daily air temperatures varied from 14 °C to 31 °C in the greenhouse during the experiment with the mean minimal value of 20.1 °C and the mean maximal value of 28.1 °C.
There were four watering frequencies (10, 6, 3 or 2 times per month) for H. ammodendron seeds from the four deserts with a monthly total rain of 50 mm, which was equivalent to 88, 147, 293 and 440 mL water each time, respectively. Seedling emergence was recorded daily for each pot, and the experiment was terminated after 30 days, at which time no seedlings had emerged for at least 5 consecutive days. All sand in each pot was passed through a sieve with 2 mm diameter to search for germinated but non-emerged seedlings, and then seed germination percentage was calculated for each treatment.
Seeds of H. ammodendron from the four deserts were tested for germination in the incubators in light (fluorescent tube with light intensity of 50 μmol m −2 s −1 ) at the Plant Ecophysiology Laboratory in Chinese Academy of Forestry. The test using four constant temperatures was conducted from 22 April to 7 May 2020; and the one using four alternating temperatures was conducted from 11 June to 26 June 2020. The period for high and low temperature was 12 h and 12 h in the alternating temperature treatments, respectively. Twenty-five seeds were uniformly placed on two layers of Whatman No. 1 filter paper and 2 mL distilled water in a 9 cm diameter plastic Petri dish using forceps. There were four Petri dishes (replicates) for each treatment and a total of 64 dishes in each experiments (4 temperatures × 4 seed sources × 4 replicates). A seed was considered to be germinated when the radicle appeared (1 mm) from the seed coat. Germination was examined daily, distilled water was added when necessary and the germinated seeds were removed from the Petri dish. The seed germination experiment was terminated after 15 days, at which time no seeds had germinated for at least 5 consecutive days.

Statistical analysis
A completely randomized design was used in all experiments. Percentages of seed germination and seedling emergence were expressed as mean ± SE (standard error). The percentages were arcsine square root transformed before analysis, but untransformed data are shown in tables and figures. Two-way MAP: mean annual precipitation; MAT: mean annual air temperature; T max : mean daily air temperature in July; T min : mean daily air temperature in January. Seed mass is the total mass of 1000 seeds. ANOVA at the 95 % probability level was conducted to compare the effects of rain frequency and seed source on the percentage of seed germination and seedling emergence of H. ammodendron, and the effects of constant temperature or alternating temperature and seed source on seed germination percentage (Sokal and Rohlf 1995). If ANOVA showed significant effects, Duncan's test was used to determine the difference between treatments. All analysis were conducted by SPSS 19.0 for Windows (SPSS Inc., Chicago, IL, USA).

Results
Effects of rain frequency on seed germination and seedling emergence Seedling emergence of H. ammodendron was delayed with decreasing rain frequency (Fig. 1). Tengger Desert seedlings of H. ammodendron emerged on the second day at the rain frequency of 6 and 3 times per month but on the fourth day at 2 times per month. Badain Jaran Desert seedlings of H. ammodendron emerged on the third day at the rain frequency of 10 and 6 times per month but on the fifth day at less frequent rain. Gurbantonggut Desert seedlings of H. ammodendron emerged on the third day at the rain frequency of 10 and 6 times per month but on the fourth day at less frequent rain. Mutthar Desert seedlings of H. ammodendron emerged on the seventh day at the rain frequency of 10 times per month but on the eighth day at less frequent rain. In general, seedling emergence of H. ammodendron was delayed for seeds from east to west, which was fastest for Tengger Desert seeds and slowest for Mutthar Desert seeds. Seed germination and seedling emergence percentage of H. ammodendron were significantly affected by rain frequency (P < 0.001), seed source (P < 0.001) and their interactions (P < 0.01, Table 4). Generally, seedling emergence declined with decreasing rain frequency (Fig. 1). The percentages of seed germination (91) and seedling emergence (89) of H. ammodendron seeds from the Tengger Desert were higher at the rain frequency of 6 times per month than those at other treatments. The percentages of seed germination and seedling emergence of H. ammodendron seeds from the Badain Jaran Desert were higher at the rain frequency of 10, 6 and 3 times per month than at 2 times per month. The percentages of seed germination (73 and 72) and seedling emergence (67 and 63) of H. ammodendron seeds from the Gurbantonggut   Desert were higher at the rain frequency of 10 and 6 times per month than those at other treatments. The percentages of seed germination (37 and 36) and seedling emergence (32 and 29) of H. ammodendron seeds from the Mutthar Desert were higher at the rain frequency of 10 and 6 times per month than those at other treatments (Figs 1 and 2). At the rain frequency of 10 times per month, the percentages of seed germination and seedling emergence of H. ammodendron were higher in Badain Jaran Desert and Gurbantonggut Desert than the other two deserts. At the rain frequency of 6 times per month, the percentages of seed germination and seedling emergence were highest in Tengger Desert. At the rain frequency of 3 times per month, the percentages of seed germination and seedling emergence were higher in Tengger Desert, Badain Jaran Desert and Gurbantonggut Desert than in the Mutthar Desert. At the rain frequency of 2 times per month, the percentages of seed germination and seedling emergence were higher in Gurbantonggut Desert and Tengger Desert than in the other two deserts (Fig. 2).

Effects of constant temperature and alternating temperature on seed germination
From the Tengger Desert, Badain Jaran Desert and Gurbantonggut Desert, seeds of H. ammodendron germinated on the second day at four constant temperatures (Fig.  3). However, seed germination of H. ammodendron from the Mutthar Desert was delayed with the decreasing constant temperature, which started on the second day at 20 °C and 25 °C but on the third day at lower constant temperatures. Seed germination percentage was affected significantly by seed source (P < 0.001) and its interactions with constant temperature (P < 0.05, Table 5). For H. ammodendron from the Tengger Desert, Badain Jaran Desert and Gurbantonggut Desert, seed germination was higher than 80 % at the four temperatures; however, seed germination of H. ammodendron seeds from the Mutthar Desert was significantly lower (75 %) at 25 °C (Figs 3 and 4). At the lower temperatures (10 °C to 20 °C), germination percentage was similar for seeds from the four deserts. However, seed germination percentage was significantly higher for Tengger Desert, Badain Jaran Desert and Gurbantonggut Desert seeds than those from the Mutthar Desert at 25 °C (P < 0.001) (Fig. 4).
From the Tengger Desert, Badain Jaran Desert and Gurbantonggut Desert, seeds of H. ammodendron germinated on the second day at four alternating temperatures (Fig.  5). However, seed germination of H. ammodendron from the Mutthar Desert was delayed with the increasing alternating temperature, which started on the second day at 20/10 °C but on the third day at higher alternating temperatures.
Seed germination percentage of H. ammodendron was affected significantly by alternating temperature, seed source and their interactions (P < 0.001, Table 5). For H. ammodendron seeds from the Tengger Desert, Badain Jaran Desert and Gurbantonggut Desert, maximum germination was 95 %, but only 78 % for seeds from the Mutthar Desert. However, seed germination of H. ammodendron from the Mutthar Desert was significantly higher (75 %) at 20/10 °C, 25/15 °C and 25/10 °C than that at 30/15 °C (36 %) (P < 0.001, Figs 4 and 5). At 20/10 °C, seed germination percentage was significantly higher in Tengger Desert and Gurbantonggut Desert than that in Mutthar Desert (P < 0.05). At 25/10 °C, seed germination percentage was significantly higher for Tengger Desert seeds than Mutthar Desert seeds (P < 0.05). At 25/15 °C and 30/15 °C, seed germination percentage was significantly higher from Tengger Desert, Badain Jaran Desert and Gurbantonggut Desert seeds than for Mutthar Desert seeds (P < 0.001) (Fig. 4). Table 4. Two-way ANOVA of response of seed germination and seedling emergence of Haloxylon ammodendron from different sources to rain frequency.  Different upper-case letters indicate significant differences between seed sources and different lower-case letters indicate significant differences between rain frequencies.

Discussion
The suitable environmental conditions for seed germination are highly unpredictable over space and time in desert ecosystems (Wang et al. 2018b). Seeds of most desert plants generally germinate after enough rainfall in the growing season, and have an opportunistic strategy to adapt to the harsh environment (Gutterman 2002). Our experiment results showed that seeds of H. ammodendron from four deserts germinated rapidly and well at the rain frequency of 10 times per month totalling 50 mm. Moreover, seed germinated well in   a relatively wide range of temperatures varying from 10 °C to 25 °C and from 20/10 °C to 30/15 °C, except the Mutthar Desert seeds that were inhibited by the highest temperature.
Response of seed germination and seedling emergence to rain frequency change Seed germination and seedling emergence of H. ammodendron from the four deserts were higher at the highest rain frequency (10 times per month) and inhibited by a decrease in rain frequency (Fig. 1). Therefore, the best time for seed germination and seedling emergence of this shrub is late spring or early summer after a few days of rainfall events. The responses of H. ammodendron seeds to rain frequency may reflect an opportunistic strategy, which would be an adaptation of the species to desert ecosystems, which is typical in many desert plants. For example, after initial irrigation of 8 mm or 16 mm, subsequent irrigation of 3 mm at 1-day or 2-day intervals resulted in high seedling emergence of three annuals (Agriophyllum squarrosum, Bassia dasyphylla and Aristida asdcensionis) in Mu Us Sandy Land, while irrigation at 4-day or 6-day intervals only favoured seedling emergence of A. adscensionis (Tobe et al. 2005b). Seedling emergence of three Artemisia species was maximal when they were initially and subsequently treated with 16 mm and 3 mm irrigation at 1-day intervals. However, when they were initially and subsequently treated with 8 mm and 3 mm irrigation at 2-day intervals, seedling emergence was almost completely suppressed due to water deficiency in sand (Tobe et al. 2006). Similarly, seedling emergence of Leymus secalinus is adapted to 150 mm of monthly rain with the frequency of 10-30 times per month, and decreased as rain frequency decreased in Mu Us Sandy Land (Zhu et al. 2014). More rains in late spring also enhanced seedling emergence of Aspidosperma quebracho-blanco, and the regular rainfall distribution rather than rainfall amount was the most significant factor in the survival of this species in central Argentina (Barchuk et al. 2005). Rain frequency was as important as rain amount to seedling emergence of R. soongarica since the highest emergence was obtained with a 30 % increase in rain amount and a 50 % reduction in rain frequency at the southern edge of the Badain Jaran Desert (Shan et al. 2018). Therefore, rain frequency plays an important role in seed germination and seedling emergence of desert plants. Less frequent rainfall will restrict seed germination and seedling emergence of desert plants in the future. In the Tengger Desert and Badain Jaran Desert, mean rain frequency is about 3 times per month during the germination season. In the Gurbantonggut Desert and Mutthar Desert, mean rain frequency is about 4 times per month during the germination season (Table 2). Under future climate change, less frequent and larger precipitation events will increase (IPCC 2013); thus, the rain regime in the arid zone would be more unpredictable. Less frequent rain events will inhibit seed germination and seedling emergence of H. ammodendron. Moreover, the regeneration of the H. ammodendron community might be more sensitive to rain change in the Tengger Desert and Badain Jaran Desert than in the Gurbantonggut Desert and Mutthar Desert, since there is less rain in the two former deserts (about 110 mm) than in the latter desert (about 150 mm) ( Table 1).

Response of seed germination to warming
Seed germination of H. ammodenron is adapted to a relatively wide range of temperatures and germination was higher at both constant (10 °C to 25 °C) and alternating temperatures (20/10 °C, 25/10 °C, 25/15 °C and 30/15 °C) for Tengger Desert, Badain Jaran Desert and Gurbantonggut Desert seeds than for Mutthar Desert seeds. Similarly, seeds of C. korshinskii from the Ordos Plateau germinated well from 5/15 °C to 25/35 °C (Lai et al. 2015). However, germination of H. ammodendron seeds from the Mutthar Desert was inhibited by high temperatures (25 °C or 30/15 °C) (Figs 3-5). Based on field observations, seed germination of H. ammodendron generally occurs in spring. The inhibition of seed germination in this area may result also from higher air temperatures in the later germination season, which were 26.82 °C and 32.44 °C in May and June, respectively (Table  3). Mean temperature for seed germination was 17.8 ± 0.6 °C in cold deserts , which was similar to spring air temperatures. High temperature in summer inhibited seed germination of many desert plants. For example, seeds of Agropyron cristatum, A. halodendron, C. korshinskii and Melilotus suaveolens from Horqin Sandy Land germinated well at 25/15 °C and 30/20 °C but poorly at 35/25 °C (Lai et al. 2016). Warming also inhibited seed germination or seedling emergence of four woody Banksia species in South Western Australia (Cochrane et al. 2015). Therefore, desert plants may germinate earlier to adapt to global warming in the future.
In deserts of Northwest China, the best time for seed germination of H. ammodendron is late spring and early summer, when temperature would be appropriate and sand would be moist after rainfall. Populations of H. ammodendron with larger seed mass may have an advantage in desert ecosystems, e.g. Tengger Desert and Gurbantonggut Desert. The effect of seed mass on seed germination and seedling emergence of H. ammodendron will need more research, especially under sand burial . It is predicted that seed germination of H. ammodendron will occur earlier under global warming. However, considering both warming and less frequent rainfall, seed germination might be more difficult for H. ammodendron, especially in the Mutthar Desert. Therefore, the regeneration of the H. ammodendron community should be enhanced in the future by irrigation and seedling transplant.

Conclusions
In the temperate desert ecosystems, H. ammodendron has an opportunistic seed germination strategy which is an adaptation to the unpredictable desert environment. Seeds of H. ammodendron from four deserts of Northwest China germinated at high rain frequencies and over a wide range of air temperatures. However, seed germination of H. ammodendron was inhibited by the highest temperature of the Mutthar Desert. Therefore, the optimal germination time for H. ammodendron is in spring and early summer after rainfall. Under future climate change, seed germination and seedling emergence of H. ammodendron would be restricted by warming and decreased frequency of rainfall in the desert ecosystem. The regeneration of H. ammodendron community should be promoted by irrigation and seedling transplant.

Supporting Information
The following additional information is available in the online version of this article. Experiment data of this research was presented in Supporting data 1 and Supporting data 2.

Contributions by the Authors
Y Zhu designed and conducted the experiment, collected seeds, analyzed data and wrote article; Z Jia applied foundation, approved experiment and revised article; G Wang revised article; H Ning, X Ji and Q Luo helped to collect seeds.