This study was conducted in the experimental area of Cele National Station of Observation and Research for Desert Grassland, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences. The experimental site is located on the southern edges of Taklimakan Desert (35°17′55″–39°30′00″N, 80°03′24″–82°10′34″E), which has a typically arid continental climate (Figure 1). The mean annual temperature is 11.9°C, ranging from 41.9°C in summer to −23.9°C in winter. The mean annual precipitation here is only 35.1 mm, and the mean potential annual evaporation is 2,600 mm (annual precipitation: annual evaporation = 0.0135) (Liu et al., 2016). Soils here are classified as aridisols (United States Department of Agriculture USDA, 1999) in the USDA ST system, with poor water holding capacity, severe desertification, and low organic matter content. The area has abundant thermal resources and more sunny days. The annual average sunshine hours are 2,854.2 h (Zhang et al., 2017). The main plant species include desert shrubs and grasses, such as K. caspia, A. sparsifolia, P. euphratica, Tamarix ramosissima, and Calligonum mongolicum, with low vegetation coverage.

Three dominant plant litter leaves of K. caspia, A. sparsifolia, and P. euphratica were collected from September to November 2017. Litter collection baskets were set under vegetation, and the baskets were 50 cm × 50 cm for K. caspia and A. sparsifolia, and 100 cm × 100 cm for P. euphratica. All samples were mixed evenly. The collected samples were dried for 48 h in a 75°C oven after the impurities were removed. Then, 20 × 20 cm nylon mesh bags with a mesh size of 1 mm were used as litter bags, and each bag was precisely filled with 20.0 g litter leaves. Six litter bags were used as a sampling unit for one species, and the size of the unit was 2 × 2 m. The whole experiment included 216 litter bags in total (3 litter species × 3 filter treatments × 6 collection times × 4 replicates).

Bags of three litter species within each treatment were respectively pinned by nail to the sand soil surface in three blocks under the following treatments:

(1) Ambient, where the bags were covered by a fully transparent film (0.05 mm plastic, China) supported by PVC pipes at an inclination angle to prevent precipitation accumulation, a 95% penetration rate of visible light (>280 nm), with data obtained by using a UV photometer (Lambda25, PerkinElmer, USA) and an LI-250A illuminometer (LI-250A, LI-COR, Lincoln, USA). PVC pipes were set 1 m on each side, and 40 cm above the soil surface to avoid potential greenhouse effects (Uselman et al., 2011) and ensure that other climate factors remain consistent for all treatments.

The optical properties of the fully transparent film, UV-shielding film, and total dark film in the study are shown in the figure (Supplementary Figure S1). All films were replaced once a month to prevent aging.

The sample plot was checked every 2 weeks. Any sand on the litter bags was removed to ensure normal experimental conditions.

Litter bags were placed in the test plot in January 2018. Four replicates of bags of each plant type/light treatment were collected from April to December 2018. The litter bags were taken back to the laboratory to dry, and the residual sand was removed with a 1 mm mesh sieve. To ensure the rationality of the experimental design, we monitored the seasonal variation of UV irradiance in the study area during the experimental year, which was higher (0.91 MJ m⁻² day⁻¹) in summer and lower (0.30 MJ m⁻² day⁻¹) in winter (Figure 2A). We also evaluated the precipitation and temperature in the study area from 2016 to 2018. The annual precipitation was 45.4, 92.4 and 53.6 mm, and the average annual temperature was 12.8°C, 12.0°C, and 11.7°C, respectively (Figure 2B).

Climatic data such as temperature, precipitation, and UV irradiance were obtained from a meteorological station located less than 50 m away from the test sample plot. Specific leaf area (surface area: mass [oven dried]) was determined by measuring the projected surface area for each litter type using a leaf area meter (S-102P, Scitek, USA). After fast and fine grinding with an MM400 grinder (RETSCH, Germany), the litter was passed through a 0.2 mm mesh. About 0.2 g of samples were weighed and wrapped in tin foil, and the total carbon content of litter was determined by an elemental analyzer (vario MACRO Cube, Germany) (50 mg phenylalanine calibration). About 0.5 g of the sample was weighed, and 10 mL concentrated sulfuric acids and a piece of catalyst were added to perform the digestion process of the sample. The Kjeldahl apparatus (FOSS, Denmark) was used to determine the nitrogen content of the litter (Ammonium sulfate with purity >99.5% was used to detect the recovery). About 0.5 g of samples were weighed to determine lignin by using a cellulose analyzer (USA Fibertec). The total amount of ash was determined by muffle furnace ashing. Carbonized litter was placed in a 550°C muffle furnace for ashing, which was used to the determine the water-soluble salt content.

The decomposition process of litter is fitted by the Olson exponential decay model (Olson, 1963), and the decomposition constant k is calculated as k = – ⁡ ln M at / M ao / t 1 where M_ao, M_at represents the initial and after decomposition time t litter’s dry weight, respectively. t₁ represents a decomposition time of 1 year.

SPSS 20.0 (IBM, USA) for Windows was used to analyze the test data. One-way ANOVA with Tukey correction was used to analyze the difference of initial litter traits on different species; the difference of mass loss rate and decomposition constant k under different illumination conditions and among various species; the difference of carbon loss, nitrogen loss, lignin loss, ash loss, and water-soluble salt loss under different illumination conditions and among various species; and the difference of mass loss, lignin loss, and carbon loss between summer and winter and under different illumination conditions. Two-way ANOVA was used to analyze the influence and interaction between treatments and species on mass loss rate, k, carbon loss, nitrogen loss, lignin loss, ash loss, and water-soluble salt loss. Before the analysis, the Shapiro–Wilk method was used to judge the normality and homogeneity of the variance of the data. The data of water-soluble salt loss rate were converted by log (10) to meet the requirements of residual normality. The test level was set as p = 0.05. Prism (GraphPad, USA) software was used for drawing plots of mass loss, decomposition constant k, carbon loss, lignin loss, and nitrogen loss under different treatments for different species, and plots of mass loss, lignin loss, and carbon loss under different illumination conditions between summer and winter for different species. Relationships among litter loss rate, temperature, precipitation, radiant energy, and geothermal energy under ambient treatment were assessed by Pearson correlations and their corresponding heatmap. A structural equation model (SEM) was established by the piecewiseSEM software package of R software (Lefcheck, 2016; Zhang et al., 2021). SEM was used to tease and quantitatively analyze the chemical components that directly and indirectly affect the mass loss of litters. D-separation test was used to check whether the model omitted important links, and p > 0.05 indicated that the model was approved (Shipley, 2002).

Significant differences were found in the initial traits of the three kinds of litter (Table 1). Concentrations of carbon and nitrogen and specific leaf area of A. sparsifolia were significantly higher than those of P. euphratica and K. caspia. The C:N ratios were highest in P. euphratica and lowest in A. sparsifolia. The lignin content of P. euphratica and A. sparsifolia was significantly higher than that of K. caspia. Concentrations of ash and water-soluble salt showed a consistent pattern, with K. caspia, A. sparsifolia, and P. euphratica representing the concentrations in descending order.

Results showed that the trend of the mass loss rate curve of the three kinds of litter was almost the same (Figure 3). After 1 year of decomposition, the mass loss rates under total solar radiation of K. caspia, A. sparsifolia, and P. euphratica were 45.4%, 39.8%, and 34.9%, respectively, and 20%, 22.2%, and 17.4%, respectively, under UV filtering treatment. Thus, UV radiation increased the mass loss of K. caspia, A. sparsifolia, and P. euphratica by 25.4%, 17.6%, and 17.5%, respectively. The mass loss rate under sunlight condition was significantly higher than that under UV filtering and dark treatments (Table 2, p < 0.05). The decomposition constant k of the three litter species under sunlight condition was significantly higher than that under UV filtering and dark treatment (Figure 4; Table 2, p < 0.05). Litter that received full sunlight had the fastest decomposition rate. However, no significant difference was found between UV filtering and dark treatments (Figure 4; Table 2, p > 0.05), UV radiation played a dominant role in photodegradation.

In addition, no significant difference was found among three species under the same illumination condition (Table 2, p > 0.05). Similarly, the difference of decomposition constant k among three species did not reach a significant level, and no interaction occurred between treatments and species on mass loss rate and k value (Table 2, p > 0.05).

Illumination treatments significantly affected the litter loss rate of carbon and lignin (Table 3, p < 0.001). The loss rates of all litter traits were highest under full sunlight. Specifically, the loss rate of carbon under full sunlight was 2.5 and 2.4 times higher than under UV filtering and dark treatments, respectively. The loss rate of lignin under full sunlight was 2.2 and 2.1 times higher than under UV filtering and dark treatments, respectively. No significant difference was found in the loss rate of carbon and lignin between UV filtering and dark treatments (Table 3, p > 0.05). The effects of solar radiation on the loss rate of ash and salt were not significant (p > 0.05), but the nitrogen content under UV filtering was significantly higher than that under ambient treatment (Table 3, p < 0.05). Similarly, the carbon and lignin of three litter showed similar results compared with the mass loss process, while the nitrogen loss was very different from the mass loss process (Figures 3, 5).

Species significantly affected the litter loss rate of nitrogen, ash, and water-soluble salt (Table 3, p < 0.001). The loss rate of ash and water-soluble salt in K. caspia was about 2.5 times than in A. sparsifolia, and more than 4 times than in P. euphratica under the same illumination condition. However they varied significantly between species, which seemed to be related to differences in initial litter traits (Tables 1, 3). In addition, species–treatment interaction was significant only for the loss of nitrogen, ash, and water-soluble salt (Table 3, p < 0.05).

Under ambient treatment, litter mass loss of all species showed a more significant difference in summer than in winter (p < 0.05). The mass loss of K. caspia, A. sparsifolia and P. euphratica increased (mass loss_summer – mass loss_winter) by 4.3%, 9.6% and 6.8%, respectively (Figure 6). In addition, the lignin and carbon loss of K. caspia and A. sparsifolia were significantly different between summer and winter (Figures 6A, B, p < 0.05). However, P. euphratica did not show seasonal differences in this regard (Figure 6C, p > 0.05). Most species loss rates did not show seasonal differences under UV filtering treatment (p > 0.05), except for mass loss of A. sparsifolia and carbon loss of P. euphratica (Figures 6B, C, p < 0.05). No seasonal differences in litter loss rates were observed for all species under dark treatment (Figure 6, p > 0.05).

In summer, all species showed significant differences in mass and lignin loss between ambient and shading treatments (p < 0.05). The mass loss of K. caspia, A. sparsifolia and P. euphratica under UV radiation increased (mass loss_ambient – mass loss_{UV filtering}) by 20.2%, 17.0% and 11.6%, respectively (Figure 6). In contrast, in winter, all species showed little difference between ambient and UV filtering treatments (Figure 6, p > 0.05). The trait loss (including mass loss, lignin loss and carbon loss) of litter was closely and positively correlated with total solar radiation, total UV radiation and optical quantum flux in summer (Figure 7A, mass loss: p < 0.001, lignin loss: p < 0.01, carbon loss: p < 0.05). However, in winter, no significant correlation existed between litter trait loss and climate factors such as temperature, precipitation, and radiation (Figure 7B, p > 0.05).

The key factors that affect the variations in litter mass loss under full sunlight treatment were assessed using SEM (Figure 8A). Under full sunlight treatment, water-soluble salt had direct effects (0.88 and 0.45) on ash and lignin, and indirect effects (0.65) on mass loss. Ash had direct effects (0.31) and indirect effects (−0.18) on mass loss. Lignin had direct effects (0.84) on mass loss. Carbon had direct effects (0.85) on lignin and indirect effects (0.71) on mass loss. Nitrogen had no effect on mass loss. Therefore, the effects of lignin on variations of litter mass loss were the highest in all traits. The standardized effect sizes of the driving factors further clarify the SEM results (Figure 8B).

Over a period of 1 year, we observed a higher loss of litter quality, ranging from 17.5% to 25.4%, in the ambient treatment compared with the treatment without UV exposure (Figure 3; Table 2). Notably, a significant difference in the impact of UV radiation on litter mass loss was observed starting from June (Figure 3, p < 0.05). When the decomposition time approached 1 year, the mean mass loss rate of litter under UV radiation was 45%, which was similar to other extremely arid deserts (Huang et al., 2017a; Becky et al., 2019). These findings align with the conclusion that photodegradation is the primary controlling factor for above-ground litter decomposition in arid ecosystems (Austin and Vivanco, 2006). The average decomposition rates of litter were 0.88, 0.38, and 0.34 yr⁻¹ under full sunlight, UV filtering, and dark treatment, respectively (Figure 4; Table 2). These rates are higher than those observed for leaf litter in previous studies, which ranged from 0.28 to 0.78 yr⁻¹ (Vanderbilt et al., 2008; Throop and Archer, 2009; Uselman et al., 2011). The difference in decomposition rates might be attributed to the intensity of UV radiation exposure to litter (Berenstecher et al., 2020). In our study area, the daily average UV radiation intensity during summer (630 KJ m⁻² day⁻¹) was much higher than that reported in other studies (Vanderbilt et al., 2008; Throop and Archer, 2009; Uselman et al., 2011), which were below 100 KJ m⁻² day⁻¹ (Figure 2A).

Surprisingly, in contrast to our first hypothesis that we did not observe any significant differences in final mass loss and decomposition rate (k) among different litter species types, regardless of the treatment (Table 2, p > 0.05). This finding is consistent with a litter decomposition experiment conducted in a semi-arid Mediterranean oak forest, where individual species did not explain the rate of litter decomposition (Sarabi et al., 2021). This result may be attributed to the fact that most of our research objects are shrubby vegetation. Furthermore, the effect of treatment–species interaction did not yield significant differences in final mass loss and decomposition rate (k) in our study area. Overall, our study demonstrated that species type had no significant effect, highlighting the importance of photodegradation in litter decomposition.

Our study demonstrated that different illumination treatments had a significant impact on specific litter trait losses in hyper-arid deserts (Table 3, p < 0.05). Specifically, UV radiation accelerated the loss of carbon and lignin in litter, which is consistent with the observed differences in litter mass loss under varying degrees of photodegradation (Tables 2, 3; Figures 5A, B). This finding aligns with previous studies that showed a preferential degradation of lignin, as a light-absorbing compound, by UV radiation (Meentemeyer, 1978; Day et al., 2007; Austin and Ballaré, 2010). In arid regions, photodegradation accelerates carbon release through direct and indirect mechanisms (Austin and Vivanco, 2006). In hyper-arid areas where plant litter is highly exposed to UV radiation, the photodegradation of lignin may directly contribute to carbon release. Furthermore, photodegradation has the potential to positively influence subsequent biological renewal by breaking down resilient cell wall polymers, such as lignin (Austin and Ballaré, 2010). We did not observe significant differences in the loss rate of carbon and lignin between the UV filtering and dark treatments, which is consistent with the mass loss findings (Table 3; Figures 5A, B, p > 0.05). In addition, no significant differences in carbon and lignin loss were among species nor did significant treatment–species interactions exist (Table 3, p > 0.05). Our results indicate that the litter traits significantly affected by UV radiation in our study area were not influenced by species.

In many ecosystems, traditional indices such as nitrogen have been used to predict litter decomposition rates (Meentemeyer, 1978; Melillo et al., 1982; Cornwell et al., 2008). However, in arid areas, these indices have sometimes been associated with decay in only a few litter types (Throop and Archer, 2009; Day et al., 2015). Similarly, these traditional indices failed to accurately predict litter decomposition in our study region. Interestingly, we found a significant difference in nitrogen loss between the ambient and UV filtering treatments, with the nitrogen loss after UV filtering treatment being significantly higher than that after ambient treatment (Table 3; Figure 5C, p < 0.05). Previous research established a positive correlation between nitrogen release and microbial activity and revealed that UV radiation accelerates nitrogen loss and negatively impacts microbial activity (Parton et al., 2007; Wang et al., 2015; Huang et al., 2017a). As indicated by a comparison of nitrogen loss under different illumination treatments, our results suggest that not only UV radiation but also other light spectrum (such as PAR) may inhibit microbial activity in hyper-arid deserts.

During summer, UV radiation led to a 20.2% increase in the mass loss rate of K. caspia, a 17.0% increase for A. sparsifolia, and an 11.6% increase for P. euphratica. In contrast, no significant difference in litter loss rate was found between treatments with and without UV radiation in winter (Figure 6, p > 0.05). This finding is consistent with our second hypothesis that UV is no longer dominant in litter decomposition in winter. These findings align with similar observations in humid areas where decomposition rates are affected by seasonal variation (Ma et al., 2017). In summer, UV radiation contributed to 9.88% of the mass loss, indicating its significant role in litter decomposition under sufficient sunlight. However, UV radiation did not significantly increase the mass loss rate in autumn and winter (Jiang et al., 2022). In addition, our study revealed that the mass loss during summer was significantly higher than that during winter, with K. caspia, A. sparsifolia, and P. euphratica experiencing increases of 4.3%, 9.6%, and 6.8%, respectively, due to seasonal variation (Figure 6). We observed that the seasonal litter photodegradation rates were significant correlated with UV radiation intensity but not with temperature or precipitation (Figure 7, p < 0.05). Notably, some studies reported higher mass loss of litter in winter than in summer possibly due to abundant rainfall (annual average of 1,100 mm) and the active role of decomposers in carbon turnover in those ecosystems (Berenstecher et al., 2020). In contrast, our study area had an average annual rainfall of only 53.6 mm (Figure 2B), which significantly limited microbial activity. Furthermore, the intensity of UV radiation in summer was higher than in winter (Figure 2A), providing a potential explanation for the higher mass loss observed during summer than in winter.

With regard to lignin loss, we found a significant difference between treatments with and without UV radiation in summer (p < 0.05), while no significant difference was observed in winter (Figure 6, p > 0.05). This finding suggests that lignin was more susceptible to UV radiation degradation in summer. These findings align with the research of Berenstecher et al. in the Patagonia woodlands, where UV radiation can break down lignin into CO₂ and small compounds through photodegradation, which is the main cause of carbon loss in this ecosystem during the summer (Berenstecher et al., 2020). Furthermore, experimental results from Austin et al. demonstrated that lignin, as an effective light-absorbing compound, is greatly influenced by light intensity and spectral range (Austin and Ballaré, 2010). In our study area, which is characterized by a hyper-arid desert environment, lignin is particularly susceptible to photodegradation due to the stronger UV radiation in summer (Figure 2A). These findings imply that the mechanism by which radiation influences litter decomposition may vary with the seasons through changes in lignin decay. Therefore, further studies are needed to explore the potential mechanisms involved.

Previous studies consistently demonstrated that lignin loss plays a significant role in mass loss during litter decomposition, particularly under the influence of photodegradation (Meentemeyer, 1978; Austin and Ballaré, 2010; Lee et al., 2012; Erdenebileg et al., 2018). These findings are in line with our own results, which indicate that photodegradation, particularly UV degradation, is the primary driver of litter decomposition in hyper-arid areas. We also observed that lignin loss had a significant impact on the final mass loss of litter (Table 3; Figure 8A). In addition, our study revealed that carbon indirectly influenced mass loss by directly affecting lignin content. Furthermore, the loss of water-soluble salts influenced the ash and lignin content, subsequently indirectly affecting the final mass loss. In contrast, nitrogen did not have a direct effect on litter mass loss. The results were similar to the findings in some arid and semi-arid areas, which is that the primary reason for greater mass loss of litter under UV radiation is photodegradation, especially the degradation of lignin (Day et al., 2007; Mendez et al., 2019). Our results provide more specific insights by highlighting the direct influence of lignin (0.84) on litter mass loss in hyper-arid deserts. These detailed findings contribute to a better understanding of decomposition processes and carbon turnover patterns in the region.