The study site is located in the southeast part (44.34° N, 87.85° E, altitude of 515 m) of the Gurbantunggut Desert (34°09′–49°08′N, 73°25′–96°24′E), northern Xinjiang, China. This desert (4.88 × 10⁴ km²) is the second largest desert in China and also a typical temperate inland desert in Central Asia. It is characterized as an area with low precipitation of 50–200 mm, mean relative humidity of < 45%, high evaporation rate (> 2000 mm), long winter, short spring, sufficient sunshine, and wide daily and annual temperature ranges (Zhang and Chen, 2002; Wang et al., 2003; Tao et al., 2013). The Gurbantunggut Desert is China’s only fixed semi-fixed desert that is affected by cold and humid air currents in the Atlantic Ocean (Qian and Wu, 2010). As a result, the climate, vegetation composition, and richness in the Gurbantunggut Desert vary similarly to that of any other desert in China. (Zhang and Chen, 2002). The desert vegetation is mainly composed of H. persicum, H. ammodendron, and small shrubs such as E. przewalskii and Artemisia songarica, as well as ephemeral and ephemeroid plants that do not exist in other temperate deserts in China (Mao and Zhang, 1994; Wang et al., 2003; Yuan and Tang, 2010). In addition, biological soil crusts are widely distributed, and they are important biological factors that maintain the stability of the surface of deserts (Mao and Zhang, 1994; Zhang and Chen, 2002; Zhang and Wang, 2008).

The assimilative branches of H. persicum, H. ammodendron, C. mongolicum, and E. przewalskii ( Figure 1 ) were collected from a long-term monitoring plot (Beishawo Plot) in later July 2019. Referring to the manual of methodological standards of Cornelissen et al., 2003, six well-grown individuals of H. persicum, H. ammodendron, and C. mongolicum species with similar height and crown size were chosen from a 100 m × 100 m sampling plot, and 10 green, mature, and fully-developed intact current-year assimilative branches from the mid and upper parts of an individual plant canopy were randomly collected in various directions for different species. For E. przewalskii, with a small body size, 30 plants of similar height and crown size were selected, and two intact current-year assimilative branches were collected from each individual. All plant samples (n = 60 for each species) were labeled and quickly placed into foam boxes with ice blocks before transferring them to the lab for analysis.

In the laboratory, the number of internodes of assimilative branches (NI) was counted, and the length of assimilative branches (LAB, cm) and also the average length of the first three internodes (i.e., located at the base, not the top) of assimilative branches (AL, cm) were measured with a ruler with the precision of 1 mm. The basal diameter of assimilative branches (BD, mm) was measured three times with a vernier caliper with the precision of 0.01 mm, and then the average BD was used. Finally, the assimilative branches were oven-dried at 70°C for 48 h to reach the constant weight, and then their dry mass (DM, g) was recorded with the scale with an accuracy of 0.1 mg. The parameters were expressed as mean ± standard error (mean ± SE).

Based on the decomposition of Type I sums of squares, the nested ANOVA was used to separately analyze the effects of different species, individuals, and assimilative branches on all traits (Nested Procedure, SAS version 8.0; SAS Institute Inc., Cary, NC, USA) (Liu et al., 2010). One-way ANOVA was used to compare the significance of differences in morphological traits of assimilative branches of the four desert plants. Levene’s test was used to test the equality of variances, and Tukey’s HSD test was used for multiple comparisons when group variances were homogeneous; Tamhane’s T₂ test was used for multiple comparisons of means with unequal variances (Búrquez et al., 2010). All parameters were expressed as mean ± standard error (mean ± Er). SPSS 19.0 (SPSS Inc. Chicago, Illinois, USA) was used for data analysis.

Model II regression analysis was employed to estimate the correlation between the traits of assimilative branches. The relationship between any two traits is represented by Y = β·X^α , where X and Y denote two parameter values, and α is the scaling exponent (West et al., 1997; Niklas, 2004; Niklas, 2005; Warton et al., 2006; Han and Fang, 2008; Smith, 2009). α = 1 indicates an isometric relationship, i.e., the dependent variable and the independent variable change uniformly or in the same proportion; α ≠ 1 represents an allometric relationship (Zhu et al., 2011). When determining allometric parameters, the power function is generally converted into the equation as follows: logY = logβ + α· logX; the reduced major axis (RMA) linear regression (i.e., Model Type II) was conducted to estimate the scaling slope, a 95% confidence interval (95% CI), and the coefficient of determination (R ²). Allometric scaling analysis was performed using the SMATR package (Falster, 2003). Origin 2019 (Originlab Corporation, Northampton, MA, USA) was used for the graphical representation of data.

C. mongolicum had a higher LAB (15.53 cm ± 2.26) than the other three species (13.46 cm ± 0.63, 12.66 cm ± 1.40, and 11.94 cm ± 0.95) (P > 0.05). ( Figure 3A ). Among the four species, H. ammodendron and H. persicum had the highest NI (19.8 ± 0.93 and 18.4 ± 1.14, respectively), while C. mongolicum and E. przewalskii had much lower NIs (6.8 ± 1.07 and 5.4 ± 0.20, respectively), and E. przewalskii had the lowest value (5.4 ± 0.20). ( Figure 3B ). The DM of H. ammodendron (0.0437 g ± 0.006) was significantly lower than that of E. przewalskii, C. mongolicum, and H. persicum (0.0766 g ± 0.006, 0.0680 g ± 0.013, and 0.0659 g ± 0.009, respectively), and no significant difference was found between the latter three species ( Figure 3C ). E. przewalskii, H. ammodendron, and C. mongolicum had the highest BD (1.39 mm ± 0.029, 1.36 mm ± 0.055, and 1.28 mm ± 0.038, respectively), with no significant difference between the three species; however, H. persicum had the lowest BD (1.13 mm ± 0.047). ( Figure 3D ). C. mongolicum and E. przewalskii had higher ALs (2.54 cm ± 0.19 and 2.64 cm ± 0.06, respectively) than the two Haloxylon species (1.06 cm ± 0.06 and 0.81 cm ± 0.02). ( Figure 3E ). In short, the assimilative branch traits of four desert shrubs differed.

There were significant positive correlations between LAB, NI, BD, and DM (P < 0.001), AL and DM (P < 0.05), and all scaling exponents were significantly less than one P < 0.001 ( Figure 4 ; Supplementary Table S1 ), indicating that the rate of linear increase in LAB, NI, BD, and AL was lower than in DM. The scaling exponent for LAB and DM of C. mongolicum had the largest value (0.832), while that of H. ammodendron had the smallest value (0.597); the scaling exponent of H. persicum and E. przewalskii exhibited moderate values, and both these species significantly differed from the other two species ( Figure 4A ; Supplementary Table S1 ). Furthermore, for LAB vs. DM, H. persicum had a higher mean Y-intercept than E. przewalskii, indicating that H. persicum had a larger LAB than E. przewalskii at given SM. The scaling exponent between NI and DM of C. mongolicum showed the largest value (0.772), followed by that of E. przewalskii (0.556), and those of H. persicum and H. ammodendron had the smallest values with no significant difference. The Y-intercept of H. ammodendron was significantly higher than that of H. persicum ( Figure 4B ; Supplementary Table S1 ), i.e., the former species produced higher NI than the latter at a given DM. The scaling exponents between BD and DM of E. przewalskii and H. persicum (0.267 and 0.258, respectively) had considerably higher values than those of C. mongolicum (0.219) and H. ammodendron (0.151), and H. ammodendron had the lowest value of scaling exponent ( Figure 4C ; Supplementary Table S1 ). In addition, the Y-intercept for BD vs. DM of E. przewalskii was higher than that of H. persicum. For AL vs. DM, the scaling exponent of C. mongolicum and H. persicum had markedly higher values (0.419 and 0.347, respectively) than that of H. ammodendron (0.273), while E. przewalskii showed no significant difference in scaling slope with the other three species ( Figure 4D ; Supplementary Table S1 ).

Similarly, significant positive correlations between NI, BD, AL, and LAB of the four shrubs (P < 0.001) were identified ( Figure 5 ; Supplementary Table S2 ). The scaling slope between NI and LAB of C. mongolicum (0.867) and E. przewalskii (0.836) did not significantly differ, but the Y-intercept of C. mongolicum was higher than that of E. przewalskii; The scaling exponents between NI and LAB of H. persicum and H. ammodendron had markedly lower values (0.661 and 0.593, respectively) than those of the other two species, and their Y-intercepts also differed ( Figure 5A ; Supplementary Table S2 ). The scaling exponents between BD and LAB of E. przewalskii, H. persicum, and H. ammodendron exhibited significantly higher values (0.401, 0.369, and 0.366, respectively) than that of C. mongolicum (0.181), and there was no significant difference in Y-intercepts between H. ammodendron and E. przewalskii, which were greater than that of H. persicum ( Figure 5B ; Supplementary Table S2 ). No significant differences in values of the scaling exponent between AL and LAB of the four shrubs were observed, but the Y-intercepts had significant differences among these shrubs ( Figure 5C ; Supplementary Table S2 ).

The scaling exponents between NI and BD of the four shrubs had significantly greater values than 1 (P < 0.001). C. mongolicum had the highest scaling slope (4.784), followed by E. przewalskii (2.084), whereas H. persicum had the lowest scaling slope (1.606). There was no significant difference in scaling exponents among E. przewalskii, H. ammodendron, and H. persicum. The Y-intercept of H. persicum was significantly higher than that of H. ammodendron, which was also significantly higher than that of E. przewalskii ( Figure 6 ; Supplementary Table S3 ). Consequently, the allometric relationships among multiple assimilative branch traits also showed strong inconsistency.

Leaf morphological traits clearly reflected the response and adaptation of plants to the environment and affected the material and energy exchanges between plants and the surrounding environment, as well as the viability, development, and change in the growth direction of plants (Smith and Geller, 1980; Lambers et al., 2008; Vogel, 2009; Li et al., 2012; Jin et al., 2019; Ren, 2022). Some studies have reported that after long-term drought stress, shrubs have evolved assimilative branches to adapt to the desert environment (Li et al., 2014; Ávila-Lovera et al., 2020). The nested ANOVAs of the present study demonstrated that the variation of NI and AL was the greatest among species, whereas the variation of LAB, BD and DM was primarily affected by the plants in the sample plot and relevant to assimilative branch collection, rather than the species.

Although H. ammodendron and H. persicum are members of the same genus, the traits of their assimilative branches differed. Figure 3 shows that H. ammodendron developed assimilative branches with shorter internodes, larger basal diameter, a greater number of nodes, and a lower dry mass than H. persicum. This morphological difference could indicate that H. ammodendron and H. persicum have evolved distinct strategies for maintaining effective photosynthesis, reducing water evaporation, and even reproducing offspring. Our findings are consistent with Zhang’s findings on the individual morphology of H. ammodendron and H. persicum (Zhang, 2002). Plants can spatially arrange their canopy structure by changing the length, diameter, and distribution of their branches, according to studies. H. ammodendron was larger and had greater branching ability and a denser canopy than H. persicum (Li et al., 2009; Wang et al., 2011; Xu et al., 2012). Therefore, we speculate that the morphological characteristics of assimilative branches of two Haloxylon species are related to the size and configuration of individual plants. The shorter the branch lengths of H. ammodendron, the more assimilative the branches, confirming Corner (1949) second rule that there was a trade-off between branch size and number (Westoby and Wright, 2003).

There was no significant difference in BD and LAB between H. ammodendron and E. przewalskii in this study, but H. ammodendron had significantly lower DM than E. przewalskii. Therefore, the findings, E. przewalskii had a lower specific leaf area (SLA) than H. ammodendron, implying that evergreen plants (E. przewalskii) have lower SLA and photosynthetic efficiency, as well as a longer leaf lifespan (Liu and Liang, 2016). The DM of H. ammodendron was lower than that of the other three shrubs, which could be attributed to the fine structure of the assimilative branches. Previous research has shown that the water storage tissues in H. ammodendron assimilative branches are well developed, accounting for more than 50% of their radius, and that the assimilative branches of this succulent plant have a very high moisture content, possibly due to a large amount of water loss after drying, resulting in a significant reduction in DM (Tan et al., 2011; Zan and Zhuang, 2017; Wei et al., 2020). E. przewalskii is an evergreen plant that has been subjected to more environmental stress than the other three species. According to studies, the dry mass of evergreens is greater than that of other deciduous plants, implying that the leaf construction cost of evergreens is frequently greater than that of deciduous plants (Niklas, 1999; Cornelissen et al., 2003; Li et al., 2008; Zhu et al., 2011).

The NI of H. ammodendron and H. persicum was significantly higher than that of C. mongolicum and E. przewalskii, as shown in Figure 3 . This difference was consistent with the individual size differences observed in four shrubs previously studied. As a result, we hypothesized that there might be a link between NI and plant individual size. Furthermore, studies have shown that the structural characteristics of leaves change with plant size, though some changes are not significant (Price et al., 2014; Huang et al., 2016; Chang et al., 2021). For leaves that initially grow between the internodes, there is a larger number of nodes per assimilative branch, which indicates that the shrub originally had more full-grown leaves. In addition, we found that the AL of H. ammodendron and H. persicum had significantly smaller values than that of C. mongolicum and E. przewalskii, and LAB and DM of H. ammodendron and H. persicum were slightly smaller than those of C. mongolicum and E. przewalskii at later growth stages. Therefore, we speculated that the length of assimilative branches developed at the earlier stages of plant growth may affect the total length of all assimilative branches, which may be related to the potential configuration of plants. For C. mongolicum and H. ammodendron with similar heights, studies have shown that the latter has the fractal dimension of the branches twice that of the former and also a more complex branch structure (Guo et al., 2015). From a biomechanical perspective, internode length, dry mass, and basal diameter may potentially play critical roles in plant resistance to bending and also in structural stability (Yang et al., 2016; Stubbs et al., 2022). Tall plants with complex branch structures not only bear greater water loss through transpiration and mechanical damage but also are more susceptible to strong wind interference. Therefore, by adjusting the length, dry mass, and basal diameter of assimilative branches, their ability to resist bending can be enhanced to some extent (Li et al., 2021).

In addition, it is undeniable that the differences in all traits of assimilative branches should not be ignored to a certain extent for the other appendages carried by assimilative branches (Li et al., 2009). Moreover, assimilative branches are not only the most important photosynthetic organ but also the basis of sexual reproduction. Since the construction and maintenance of tips, stipules, and other appendages undoubtedly require biomass allocation, the existence of these structures may affect the existence of assimilative branches and cause changes in them, just as the fruit structure affects the cross-species variation in seed yield (Henery and Westoby, 2001). However, it must be pointed out that the four shrubs can grow and develop normally in the arid environment due to their multiple channels and successful adaptation to drought, and any adaptation mode has its value (Zhang, 2002).

The allometric relationship among different functional traits is an ecological strategy for plants to promote resource utilization which reflects the ability of plants to coordinately adapt to environmental changes (Wu et al., 2021). The results of a goodness-of-fit test for all morphological traits of assimilative branches showed a good fit, and all traits showed a non-proportional increasing trend, indicating that the allocation ratio of resources among assimilating branch traits was different. Since DM, NI, and LAB of assimilative branches are the material basis of plant growth and reproduction, as well as the key factor for morphogenesis of photosynthetic organs, resources are more likely to be allocated to them, similar to the law of “diminishing returns” of investment supported by Niklas (Niklas et al., 2007). The allometric relationship among morphological traits of assimilative branches had specificity, indicating that different species have adapted to environmental conditions in different types of the morphological structure of their leaves (Yang et al., 2004; Guerra and Scremin-Dias, 2018). It might be due to differences in fine structures (palisade tissue, stomatal structure, etc.) of assimilative branches.

We found that the scaling exponent between all traits and the DM of assimilative branches of H. persicum also were higher or significantly higher than those of H. ammodendron, and there were no significant differences in the scaling exponent between NI, BD, and AL and between LAB, NI, and BD. The habitat heterogeneity and different ecological adaptation strategies of plants lead to differences in investment trade-off mechanisms, which directly or indirectly affect the growth, development, and reproduction of plants (Xu et al., 2014; Wu et al., 2021).

The scaling exponents between LAB, NI, AL, and DM, as well as between NI and LAB and NI and BD in assimilative branches of C. mongolicum were higher than those of the other three shrubs, while the scaling exponent between BD, DM, and LAB was the smallest. The results indicated that compared with the other three shrubs in C. mongolicum, more resources were allocated for the length and number of nodes of assimilative branches but less for their diameter. This may be related to the clonal growth of C. mongolicum (Li SY et al., 2013; Fan et al., 2016). Studies have reported that the assimilative branches of C. mongolicum show high phenotypic plasticity when exposed to heterogeneous environmental conditions such as wind erosion and sand burial (Balestri and Lardicci, 2013; Fan et al., 2016; Fan et al., 2018a; Fan et al., 2018b). Clonal plants acquire resources in the micro-environment through morphological changes and biomass allocation; for example, to escape from unfavorable habitat conditions as soon as possible and reduce their losses, they mainly establish the simplified structure and slow down their growth by developing thinner stolons, increasing the length of internodes, and reducing the number of branches (D’Hertefeldt and van der Putten, 1998; Zhang et al., 2022).

In addition, the allometric relationships between some traits of E. przewalskii were not significantly different from those of the other three plants. Among them, the assimilative branches of E. przewalskii and H. ammodendron had the same allometric relationship between their BD and LAB, which indicated that although there are quite phylogenetic and individual differences between them, they have experienced the same geological and historical events and the arid and seasonal temperate desert climate during their evolution (Xie et al., 2012) and the basal diameter and total length of their assimilative branches have a relatively consistent resource allocation rate.

However, E. przewalskii, an evergreen shrub, is better able to cope with environmental stress than the other three deciduous plants and prefers to allocate limited photosynthetic products for leaf thickening to resist harsh environments during the non-growing seasons and improve its stress resistance (Cornelissen et al., 2003; Wei et al., 2020). In addition, several reports have shown that evergreen plants take a long time to establish the supporting structures (e. g., basal diameter), which may decrease their photosynthetic efficiency so that they accumulate photosynthetic products by extending the lifespan of their leaves and increasing the length of time during which photosynthesis occurs in leaves (Westoby and Wright, 2003; Wright et al., 2004; Li et al., 2008; Choat et al., 2012).

Under the restriction of resources, natural selection develops the final geometry and structure of plant leaves in the most economical way. Only by constantly adjusting the proportion of distributed internal resources and ultimately establishing a morphological structure conducive to their growth can plants enhance their adaptation to adversity (Yang et al., 2004). As the important organ of the desert shrub, the complexity of morphology and internal structure of assimilative branches is closely related to the environment. During the long-term evolution, under the influence of systematic classification, genetic development, and environmental factors, each species has its unique biological characteristics and evolutionary adaptation process, and there are differences in the response and adaptation to the external environment; thus, different species have the same or different allometric relationships among their various traits (Liu et al., 2008). However, there are many similar adaptation strategies. Data obtained on the morphological and structural parameters of assimilating branches of four shrubs are limited, and therefore, more data need to be collected to verify the specific reasons for these similarities and differences.