The relationship between species evolution and protein evolution has been remaining as a mystery. The recent development of artificial intelligence provides us with new and powerful tools for studying the evolution of proteins and species. In this work, based on the AlphaFold Protein Structure Database (AlphaFold DB), we perform comparative analyses of the protein structures of different species. The statistics of AlphaFold-predicted structures show that, as species evolve from prokaryotes to eukaryotes, from unicellular to multicellular organisms, from invertebrates to vertebrates, and so on, the proteins within them evolve towards larger radii of gyration, higher coil fractions, higher modularity, and slower relaxations, indicating that the average flexibility of proteins gradually increases in the species evolution. With the scaling analyses, the size dependence of proteins' shape, topology, and dynamics suggest a decreasing fractal dimension of the proteins in species evolution. This evolutionary trend is accompanied by the increasing eigengaps in the vibration spectra of proteins, implying that the proteins are statistically evolving towards higher functional specificity and lower dimensionality in the equilibrium dynamics. Furthermore, we also uncover the topology and sequence bases of this evolutionary trend. The residue contact networks of the proteins are evolving towards higher assortativity, and the hydrophobic and hydrophilic amino acid residues are evolving to have increasing segregations in the sequences. This study provides new insights into how the diversity in the functionality of the proteins increases and their plasticity grows in evolution. The evolutionary laws implied by these statistical results may also shed light on the study of protein design.