One of the most important foundations of chemical geodynamics is that isotope anomalies of radiogenic isotopes observed in mantle-derived samples after correcting for natural and instrumental mass-dependent fractionations are mainly caused by radioactive decay of reservoirs with different parent-daughter ratios. It often denies the possibility of mass-independent fractionation processes at high temperatures. Therefore, isotope anomalies with very small magnitudes (ppm-level), such as W-182, has been used to identify different mantle reservoirs, the time scales of their formation and potential core-mantle interactions. W-182 anomalies are generally considered as the sole consequences of radioactivity and nucleosynthesis that could be explained via simple mixing models of multiple reservoirs. However, the nuclear field shift effect (NFSE), has proved to be capable of producing the "anomalous mass effect" especially for heavy metal isotope systems even under very high temperatures. Therefore, it is necessary to test whether such small NFSE-induced mass-independent isotope fractionations can be magnified during unique mantle evolutionary processes, such as multi-stage melting and crystallization, to produce the observed isotope anomalies in mantle-derived rocks. Here we design a multistage closed-system melting and crystallization evolution model (denoted as MC2-model), combined with ab-initio calculations and Monte Carlo simulations to test our hypothesis. Multi-stage melting and crystallization evolution can occur in magma chambers, during tectonic movements in the early earth, in complex partial melting processes or plume and its surrounding mantle. Our simulation results show that there is an amplification effect during such multi-stage evolution process. The final isotope fractionations are scaled as aN, where N is the total number of melting or crystallization processes and a is a factor that related to the evolution path and detailed melting or crystallization behaviors, such as partition coefficient (D) and degree of melting or crystallization (F). In other words, if a mantle source region experienced multi-stage melting, melt extraction and crystallization processes, the isotope effect will probably linearly magnified. Taking O and W isotopes as examples, we conduct a statistical analysis for the results of such multi-stage simulation experiments and concluded that some of the ppm-level W-182 anomalies, both positive and negative observed in Archean mantle-derived and modern plume-derived samples might be explained by this way, but our model seems to have difficulty in explaining O-17 anomalies observed in anorthosite and basalts. This study provides another perspective for the origin of isotope anomalies that are observed in mantle-derived samples. (C) 2020 Elsevier Ltd. All rights reserved.