Diverse and abundant siliceous sponge spicules were found in the latest Permian beds, Dongpan andMa'anying sections, South China, including 52 types and 85 forms. Further investigation on these spicules allows us to understand extinction patterns and processes of deep-water sponges. These sponge spicules rapidly decreased below the Permian/Triassic boundary (PTB), and the extinction rates reach up to 88%-90% for types and 88%-92% for forms. Their extinction pattern is a gradual one that consists of two stages: the first is characterized by a gentle and slow extinction speed and low extinction rate, and the second by sharp and fast extinction speed and high extinction rate. The morphological extinction process is involved in the disappearance first of the triaxons and tetraxons, then of the polyaxons and demas, and last of monaxons. In exterior structure extinction, the complex spicules with branches and spines became extinct more easily than did smooth spicules. After the end-Permian mass extinction, only five common and smooth forms survived: Oxeas A, Oxeas B, Strongles B, Oxy-orthpentactines and Oxy-orthohexactines A.siliceous sponge spicules, Permian-Triassic extinction, pattern, process, South ChinaThe end-Permian mass extinction is the largest event among the "Big Five" during the Phanerozoic history in terms of the severity of taxonomic diversity losses [1,2] . Marine animals largely disappeared, resulting in about 51%-57% families extinction, 83%-84% genus extinction and more than 95% species extinction [3][4][5][6][7][8] . This mass extinction was finely recorded in South China, thanks to the well-preserved marine strata (from shallow to abyssal). The end-Permian extinction patterns and processes of marine animals have been widely reported, and some paleobiologists thought that this extinction is sudden with only one crisis stage [9][10][11][12][13] , while the others suggested the extinction was gradual and consisted of several crisis stages [14,15] .The fossil rate is affected by biotic shape, abundance, ecological behavior, distribution and average age, and others. The first and last occurrences of one species are rarely recorded in fossils and most fossils disappeared before their extinctions. Therefore, the time recorded by fossils does not necessarily represent the real living time of biota. In other words, the biotic extinction pattern is not always represented by the fossil extinction pattern [16] . As to biologic extinction patterns, Signor and Lipps [17] showed that a random distribution of errors at the end points of biostratigraphic ranges may produce range truncations and appear gradual decline preceding a sudden extinction boundary. This phenomenon had been known as the "Signor-Lipps effect" [18,19] . Meldahl [20] empirically confirmed the "Signor-Lipps effect", noting