Spatiotemporal Dynamics and Influencing Factors of Ecosystem Carbon Use Efficiency in Central Asia

Ecosystem carbon use efficiency (CUE = NEP/GPP) is an indicator of an ecosystem’s capacity to convert photosynthetically fixed carbon into net carbon storage. Under global warming and CO2 fertilization, vegetation greening may be accompanied by asynchronous changes in carbon use efficiency. As the world’s largest non-zonal arid region, the arid region of Central Asia still lacks systematic understanding of the spatiotemporal evolution of CUE and its primary associated factors. This study analyzed the spatiotemporal dynamics and influencing factors of CUE in the arid region of Central Asia from 1999 to 2019. The results showed that the region as a whole acted as a carbon sink, with multi-year mean GPP and NEP of 414.8 g C m−2 and 83.3 g C m−2, respectively, and interannual growth rates of 6.5 g C m−2 yr−1 and 1.3 g C m−2 yr−1, respectively. The mean CUE was 0.17, and its trend was essentially stable. Shrubland (SL) had the highest CUE (0.25) but showed a slight declining trend, suggesting that its carbon use efficiency may face a risk of decline. Partial correlation analysis identified the primary factors associated with CUE in Central Asia from 1999 to 2019. Across the entire region, LAI had the highest proportion as the primary associated factor. Forest (FR) had a relatively high proportion of pixels with TEM as the primary associated factor. SL showed relatively balanced associations with multiple moisture factors, including PRE, SM, and VPD. Grassland (GL) and sparse vegetation (SV) had VPD and LAI as their primary associated factors, while in cropland (CL), LAI was prominent as the primary associated factor. Geographical detector results indicated that the statistical explanatory power of multi-factor interaction combinations for the spatial differentiation of CUE was higher than that of single factors, and two-factor interactions significantly enhanced explanatory power. Across the entire region, TEM∩LAI had the highest explanatory power (q = 0.252); some combinations exhibited bi-factor enhancement, while the rest mainly showed nonlinear enhancement. Among different ecosystems, TEM∩LAI and VPD∩LAI in SL had the highest explanatory power (q = 0.501 and 0.510, respectively), TEM∩PRE in GL had prominent explanatory power (q = 0.324), and SV and CL exhibited strong multi-factor interaction characteristics. These results reveal the spatial heterogeneity of carbon use efficiency in the arid region of Central Asia and the explanatory power of its associated factors, and further highlight that high productivity does not necessarily correspond to high carbon use efficiency, providing new insights into understanding ecosystem carbon use strategies in arid regions.