Crustal evolution events in the Chinese continent: evidence from a zircon U-Pb database

ABSTRACT

The zircon U-Pb chronology database provides a good opportunity to obtain important zircon growth peak periods in the Earth's history so as to study the origin and evolution of the crust. It should be noted that research preference affects the objectivity of zircon sampling, leading to hot data in the database and age statistics. To evaluate the influence of hot data on statistical results, the W and Y indexes are introduced. Using a Gaussian model of multipeak fitting of zircon U-Pb age frequencies, we identify seven major growth peaks in zircons from the Chinese continental crust, which are 2498.95, 1855.82, 828.88, 444.29, 249.46, 131.96, and 58.21 Ma. Due to differences in the time scales of zircon growth peaks, these peaks can be divided into two categories: first-order zircon growth peaks (I) and second-order zircon growth peaks (II), which represent longer and shorter time scales, perhaps due to different kinds of geological dynamics, respectively. In addition, there are clear correspondences between these ages and various geological events recognized by most scholars, namely, the Wutai orogeny, Lvliang orogeny, Jinning orogeny, Caledonian orogeny, Indosinian orogeny, Yanshanian orogeny, and Himalayan orogeny, respectively.

KEYWORDS:

• Crustal evolution
• major geological event
• time scale
• zircon U-Pb chronology database
• hot data

1. Introduction

Zircon occurs in trace amounts in a wide variety of quartz-bearing igneous rocks. It is physically and chemically robust under a wide range of geological pressures and temperatures, so it is also common in metamorphic rocks and as a detrital mineral in sediments and sedimentary rocks. Crystallizing zircon incorporates a variety of trace elements, including Hf, the rare earth elements, U, and Th. Because it incorporates little Pb and has a high closure temperature for trace element diffusion, zircon is widely used for geochronology using the radioactive decay of U and Th to Pb (Williams 2015). Since zircon can survive the partial melting of its host rock (Williams 2015), it is not unusual for zircon grains from igneous or metamorphic rocks to undergo a complex history of magmatic activity and metamorphic events. In addition, the high-resolution ion probes makes it possible to study individual grains containing several generations of growth (Davis, Williams, and Krogh 2003; Pell, Williams, and Chivas 1997; Armstrong et al. 1990; Kalsbeek and Nutman 1996; Williams 1992). Therefore, from the perspective of zircon growth, we can digitize the solid Earth and use the methods of big data to discuss the episodes of large-scale magmatic activity and metamorphic events in geological history. Geochronological big data not only make geochronology more quantitative but also provide chances to study it over time, space and higher dimensions (Puetz et al. 2018; Fang et al. 2018; Wu et al. 2019). In addition, scientific understanding beyond the meaning of geochronology itself can be drawn, and a new understanding of geodynamics and earth system science can be obtained through data analysis.

Although the zircon U-Pb geochronology database is of great significance to the study of geoscience, the existence of hot data, which originates in excessive research on some specific areas or topics, affects the objectivity of the database to a certain extent. Ideally, to calculate the U-Pb age distribution of zircons in the continental crust, simple random sampling of zircons in the continental crust should be performed (Puetz et al. 2018). However, under practical research conditions, random sampling is difficult to achieve. Especially in some cases, researchers have been more inclined to study areas of research focus, namely, areas that are popular among researchers and prone to over-sampling. In addition, researchers typically gather zircons by a combination of convenience sampling (predetermined site selection based on specific geological questions under investigation) and cluster sampling (collecting a cluster of zircons at the site, rather than a single zircon grain) (Puetz et al. 2018). Therefore, hot data are inevitable in the U-Pb geochronology database. However, hot data caused by nonrandom samples increase the likelihood of biased estimates (Etikan, Musa, and Alkassim 2016).

On the basis of previous research, further statistical analysis and processing of zircon U-Pb ages from the perspective of time are conducted in this paper, which is organized as follows: in Section 2 we describe the database used in this research and the methods applied on the age error fitting and Gaussian multipeak fitting. Section 3 presents the results of the computations. Section 4 discusses measuring the impact of hot data, different time scales of zircon growth peaks, zircon growth of various lithologies and the correspondence between zircon growth peaks and geological events. Finally, in Section 5 we summarize our work and draw conclusions.

2. Data and methods

2.1. Data

The data used in this study are obtained from a database that houses single-grain zircon U-Pb geochronological data for the Chinese continental crust (Fang et al. 2018; Wu et al. 2019). The database has 543,978 geochronological records and incorporates its data from more 10,000 journals accessed through the China National Knowledge Infrastructure (CNKI) and Elsevier Science, covering to the end of 2017. The geochronological data within the database comprise eight major data categories: data number, element and isotope contents, isotope ratios and their errors, ages and their errors, experimental methods, sample background, quality control, and other supplementary data. According to the lithology of the zircon host rock, the U-Pb chronological data in the database can be divided into three categories: sedimentary rock, igneous rock and metamorphic rock. However, the host rocks of some U-Pb samples are unknown. The database records are segregated into four lithology classifications: 164,845 sedimentary, 232,747 igneous, 123,594 metamorphic (including gneiss), and 22,792 with unknown lithologies.

2.2. Methods

2.2.1. Age error fitting

The first step was to obtain concordant ages from the database. Concordant ages were evaluated from ages derived from ²⁰⁶Pb/²³⁸U, ²⁰⁷Pb/²³⁵U and ²⁰⁷Pb/²⁰⁶Pb ratios. If the age data for a given sample point had all three age types and these three values were consistent within a 1σ absolute error range (or the sample point had only two age types, with both values within a 1σ absolute error range), then they were considered concordant ages (Spencer, Kirkland, and Taylor 2016).

The second step calculated the fitting curves of age errors. Absolute errors of different ages were calculated on the basis of the concordant ages. To make statistical age errors close to real errors, the idea of the Riemann sum (Guillemin 1997; Philippe and Robert 2001) was introduced into the calculation of age errors. The time-axis resolution was set to 1 Ma in this research. Then, mean age errors in each age interval were calculated to obtain error curves that changed with age. Considering the various chronometers, we performed separate statistical analyses on the ages from ²⁰⁶Pb/²³⁸U, ²⁰⁷Pb/²³⁵U and ²⁰⁷Pb/²⁰⁶Pb ratios, before which abnormal data were removed using the box plot method (Williamson, Parker, and Kendrick 1989). For each age interval, an expected value for the age errors in each interval was calculated and assigned its median value, yielding a series of age-error data. Next, scatter plots were drawn with the absolute error of age plotted along the vertical axis (y) and age along the horizontal axis (x). These scatter plots were fitted using various functions. The BoxLucas1 model (Fang et al. 2018; Wu et al. 2019) was used to fit the ages derived from ²⁰⁶Pb/²³⁸U and ²⁰⁷Pb/²³⁵U ratios:(1) $y=a\ast [1-\mathrm{e}\mathrm{x}\mathrm{p}(-b\ast x\left)\right]$(1)

In contrast, the fitting function for the ²⁰⁷Pb/²⁰⁶Pb age was the Log3P1 model (Fang et al. 2018; Wu et al. 2019), given by:(2) $y=a-b\ast \mathrm{l}\mathrm{n}\left(x+c\right)$(2)

The third step was to predict age errors. Based on the fitting curves of age errors, the prediction interval of age error at the 95% confidence level was calculated. The fitting results of each curve are shown in Table 1. Scatter diagrams of age errors, as well as fitting curves and prediction interval zones, are shown in Figure 1.

Figure 1. Absolute age error scatter plots, error fitting curves and prediction interval zones at the 95% confidence level: (a) ²⁰⁶Pb/²³⁸U ages; (b) ²⁰⁷Pb/²³⁵U ages; (c) ²⁰⁷Pb/²⁰⁶Pb ages.

Table 1. Error fitting curves for different age types.

Dependent variable	Regression equation	R-squared	Parameter	value
Age (^206Pb/^238Th)	y = a(1– e^-bx)	91.93%	a	68.14
			b	2.07E-4
Age (^207Pb/^235Th)	y = a(1– e^-bx)	71.28%	a	17.83
			b	2.29E-3
Age (^207Pb/^206Pb)	y = a-bln(x + c)	90.30%	a	216.71
			b	24.84
			c	26.03

In the final step, the recommended ages were derived based on the statistic results of the database. By comparing the absolute error curves of all three age types, we could obtain ages with the smallest errors in different times, that is, the recommended age. According to the fitting function of the error curves, intersection points were calculated to obtain the thresholds of the recommended ages. The ages derived from ²⁰⁶Pb/²³⁸U, ²⁰⁷Pb/²³⁵U and ²⁰⁷Pb/²⁰⁶Pb ratios were the recommended ages for zircons younger than 1394.64 Ma, between 1394.64 and 2994.46 Ma and older than 2994.46 Ma, respectively. Since the majority of zircon ages are younger than 2994.46 Ma in the database, only ²⁰⁶Pb/²³⁸U and ²⁰⁷Pb/²³⁵U ages were included in subsequent statistics.

2.2.2. Gaussian multipeak fitting

Concordant ages were binned with 5 Ma resolution and frequencies of ²⁰⁶Pb/²³⁸U, ²⁰⁷Pb/²³⁵U and ²⁰⁷Pb/²⁰⁶Pb ages were counted, and Gaussian multipeak fitting (Guo 2011) was carried out to explore the characteristics of zircon growth in geological history. The Gauss model in the Origin software was used for Gaussian multipeak fitting, and the specific equation is as follows: (3) $y=y_0+A/\left[w\ast {\left(\pi /2\right)}^{1/2}\right]\ast \exp \left[-2\ast {\left(x-x_c\right)}^2/w^2\right]$(3) where y₀ is offset, A is area, w is width and x_c is center.

The Levenberg–Marquardt optimization algorithm (Singh, Gupta, and Gupta 2007) was used for Gaussian multipeak fitting, which might obtain local optimum results. Considering this factor, the initial values of Gaussian peaks were set before implementing fitting depending on the visualization of the age frequency histogram so that the global optimum was obtained.

The mean value (x_c), standard deviation (σ), FWHM (full width at half maxima) and height of each single Gaussian peak were obtained by Gaussian multipeak fitting. To measure the relationship between the standard deviation of the Gaussian peak (${\sigma }_G$) and that of the experiment (${\sigma }_L$), the W index was defined as follows:(4) $W={\sigma }_G/{\sigma }_L$(4)

In the formula above, the experimental standard deviation (${\sigma }_L$) represents the upper limit of the prediction interval of the absolute age error fitting curve at the 95% confidence level. To measure the aspect ratio of a Gauss peak, the Y index was defined as follows:(5) $Y=\mathrm{F}\mathrm{W}\mathrm{H}\mathrm{M}/\mathrm{H}\mathrm{e}\mathrm{i}\mathrm{g}\mathrm{h}\mathrm{t}$(5)

To fully study peak periods of zircon growth in geological history, this paper presents Gaussian multipeak fitting for the zircon U-Pb geochronology data of all lithologies and of single lithologies (sedimentary rock, igneous rock and metamorphic rock). To analyze zircon growth at different time scales, the first step was to apply Gaussian multipeak fitting to data for the whole geological history to identify great zircon growth peaks on a long-term scale. The second step was to research zircon growth in detail on a shorter time scale, dividing geological history into four stages, namely, stage IV, stage III, stage II and stage I. The Section 4.3 results identify these temporal stages. Then, Gaussian multipeak fitting was applied to the data of each stage, after which a series of peaks were obtained, which represented secondary geological events within great peaks.

3. Results

3.1. All lithologies

Age frequency statistics and Gaussian multipeak fitting were conducted on ²⁰⁶Pb/²³⁸U and ²⁰⁷Pb/²³⁵U ages of all lithologies, which could obtain the mean value, standard deviation, W index (Equation (4)), and Y index (Equation (5)) of zircon growth peaks (Gaussian peaks), as well as the upper limit of prediction interval for the absolute error of the mean value at the 95% confidence level. The results are shown in Tables 2 and 3. What should be noted is the criteria of the concordant ages, the concordance of which is for each sample test point, i.e. each data record in the database. Considering this, when hundreds of thousands of records are calculated for statistics, results of different age types are unnecessary to be the same. However, it is apparent that the peak periods of zircon growth obtained from ²⁰⁶Pb/²³⁸U and ²⁰⁷Pb/²³⁵U ages are almost the same in this study.

Table 2. Zircon growth peaks of ²⁰⁶Pb/²³⁸U ages.

Lithology	Stage	Great geological events in the whole Earth history					Detailed geological events of stages
		xc/Ma	σ/Ma	Upper limit of age error prediction/Ma	W	Y	xc/Ma	σ/Ma	Upper limit of age error prediction/Ma	W	Y
All lithologies	Stage IV						25.92	8.73	5.86	1.49	0.03
		58.21	41.06	6.32	6.50	0.11	51.93	6.71	6.23	1.08	0.02
							130.85	4.25	7.32	0.58	0.00
		131.96	18.93	7.34	2.58	0.01	131.70	29.59	7.33	4.04	0.03
							221.13	8.96	8.55	1.05	0.01
		249.46	54.07	8.93	6.05	0.06	248.16	7.62	8.91	0.86	0.02
							274.76	50.75	9.27	5.47	0.08
		444.29	41.88	11.49	3.64	0.06	437.11	21.26	11.40	1.87	0.03
							497.02	18.72	12.17	1.54	0.06
	Stage III	828.88	106.77	16.25	6.57	0.38	809.93	53.30	16.03	3.33	0.15
							915.65	15.09	17.27	0.87	0.18
							963.56	31.57	17.83	1.77	0.31
							1074.98	78.25	19.11	4.10	1.76
	Stage II						1604.68	23.26	24.77	0.94	1.86
							1729.82	43.85	26.02	1.68	1.29
		1850.04	103.79	27.20	3.82	0.86	1847.85	30.12	27.17	1.11	0.28
							1895.06	87.88	27.63	3.18	1.63
	Stage I	2492.74	60.37	32.99	1.83	0.66	2492.76	60.13	32.99	1.82	0.66
Sedimentary rocks	Stage IV						23.24	0.15	5.83	0.03	0.00
		51.27	5.50	6.22	0.88	0.03	52.46	6.05	6.24	0.97	0.04
							94.33	7.93	6.82	1.16	0.04
		111.79	30.78	7.06	4.36	0.14	119.36	9.71	7.16	1.35	0.09
							133.87	29.90	7.36	4.06	0.35
		265.87	46.19	9.15	5.05	0.21	249.17	15.95	8.93	1.79	0.08
							300.64	19.04	9.61	1.98	0.23
							300.64	72.39	9.61	7.53	1.37
		446.46	43.22	11.52	3.75	0.20	437.82	14.85	11.41	1.30	0.08
							477.30	50.06	11.92	4.20	0.57
	Stage III	855.02	163.23	16.56	9.86	1.58	814.78	55.84	16.08	3.47	0.41
							916.15	13.21	17.28	0.76	0.49
							964.33	33.34	17.84	1.87	0.54
							1086.15	87.25	19.23	4.54	2.75
	Stage II						1432.37	22.61	23.00	0.98	2.21
							1614.72	72.75	24.87	2.92	6.25
							1734.10	34.45	26.07	1.32	1.36
		1813.80	151.00	26.85	5.62	3.66	1839.08	41.50	27.09	1.53	0.88
							1957.05	85.29	28.21	3.02	6.31
	Stage I	2474.86	75.95	32.84	2.31	2.80	2475.35	70.06	32.84	2.13	2.62
Igneous rocks	Stage IV						39.40	23.63	6.05	3.90	0.12
		132.49	91.07	7.34	12.40	0.28	130.76	4.12	7.32	0.56	0.00
		132.49	12.81	7.34	1.74	0.01	135.40	26.88	7.38	3.64	0.03
							220.41	9.50	8.54	1.11	0.02
		259.93	56.67	9.07	6.25	0.15	248.02	4.96	8.91	0.56	0.02
							270.98	48.80	9.22	5.29	0.13
		439.85	32.18	11.43	2.81	0.09	438.46	25.96	11.42	2.27	0.07
	Stage III						635.87	20.09	13.91	1.44	1.28
							782.22	44.96	15.69	2.86	0.71
		812.13	48.01	16.05	2.99	0.38	824.13	14.51	16.20	0.90	0.14
							867.85	78.43	16.71	4.69	2.73
	Stage II						1436.97	2.91	23.05	0.13	0.14
							1785.29	120.20	26.57	4.52	12.25
		1850.95	32.28	27.20	1.19	0.84	1852.54	17.70	27.22	0.65	0.50
							1976.17	179.35	28.39	6.32	40.87
	Stage I	2487.56	41.05	32.94	1.25	2.85	2486.56	48.75	32.93	1.48	3.03
							2651.74	34.17	34.30	1.00	9.92
Metamorphic rocks	Stage IV	34.82	12.26	5.99	2.05	0.16	33.26	8.46	5.97	1.42	0.13
		129.51	32.29	7.30	4.42	0.58	127.61	14.06	7.28	1.93	0.31
		234.70	22.77	8.73	2.61	0.17	223.65	7.98	8.58	0.93	0.07
							246.93	5.65	8.90	0.63	0.06
		306.97	20.64	9.70	2.13	0.37	267.14	59.06	9.17	6.44	1.66
							436.16	19.99	11.39	1.76	0.15
		457.85	51.57	11.67	4.42	0.39	494.67	15.99	12.14	1.32	0.14
	Stage III	831.81	109.35	16.29	6.71	1.44	797.44	56.13	15.88	3.54	0.72
							932.91	47.50	17.48	2.72	1.20
	Stage II	1875.79	79.26	27.44	2.89	1.33	1846.33	28.99	27.16	1.07	0.83
							1897.32	73.83	27.65	2.67	1.99
							2149.52	49.15	29.99	1.64	8.11
	Stage I	2502.67	58.75	33.07	1.78	1.21	2502.83	53.62	33.07	1.62	1.11

Table 3. Zircon growth peaks of ²⁰⁷Pb/²³⁵U ages.

Lithology	Stage	Great geological events in the whole Earth history					Detailed geological events of stages
		xc/Ma	σ/Ma	Upper limit of age error prediction/Ma	W	Y	xc/Ma	σ/Ma	Upper limit of age error prediction/Ma	W	Y
All lithologies	Stage IV						26.08	8.11	5.40	1.50	0.03
		57.46	37.38	6.57	5.69	0.10	52.96	7.29	6.41	1.14	0.02
							131.80	4.72	9.02	0.52	0.01
		132.63	19.92	9.04	2.20	0.01	132.55	29.31	9.04	3.24	0.03
							222.71	10.48	11.50	0.91	0.02
		250.11	54.28	12.15	4.47	0.06	248.76	6.70	12.12	0.55	0.02
							274.62	50.25	12.70	3.96	0.09
		445.57	42.32	15.78	2.68	0.06	438.11	21.99	15.67	1.40	0.03
							498.82	18.84	16.51	1.14	0.07
	Stage III	835.28	115.62	19.57	5.91	0.46	813.52	54.92	19.43	2.83	0.18
							951.87	37.58	20.18	1.86	0.35
							1069.97	84.41	20.66	4.09	1.98
	Stage II						1595.76	29.08	21.74	1.34	2.86
							1720.42	36.80	21.85	1.68	2.42
		1855.82	107.93	21.94	4.92	1.07	1852.82	25.28	21.94	1.15	0.29
							1877.04	102.24	21.95	4.66	1.98
	Stage I	2498.95	49.51	22.14	2.24	0.54	2499.17	47.60	22.14	2.15	0.52
Sedimentary rocks	Stage IV	52.81	6.10	6.40	0.95	0.04	54.34	6.34	6.45	0.98	0.04
							96.43	9.19	7.90	1.16	0.05
		113.01	30.23	8.43	3.58	0.14	124.06	11.82	8.78	1.35	0.07
							157.53	11.15	9.77	1.14	0.16
		265.45	46.84	12.49	3.75	0.22	246.62	11.68	12.07	0.97	0.09
							280.17	47.72	12.82	3.72	0.38
		448.60	44.59	15.82	2.82	0.22	439.64	19.84	15.69	1.26	0.09
							502.44	26.87	16.56	1.62	0.36
	Stage III	858.50	163.17	19.70	8.28	1.61	816.41	57.34	19.45	2.95	0.44
							959.88	39.32	20.22	1.94	0.58
							1096.56	82.11	20.75	3.96	2.65
	Stage II						1435.07	29.86	21.53	1.39	3.36
							1596.69	63.29	21.74	2.91	5.38
							1743.22	47.39	21.87	2.17	2.73
		1822.48	158.43	21.92	7.23	4.48	1846.06	33.06	21.94	1.51	0.88
							1923.01	113.63	21.98	5.17	7.13
	Stage I	2482.42	63.99	22.14	2.89	2.34	2483.30	57.42	22.14	2.59	2.14
Igneous rocks	Stage IV	67.36	51.99	6.92	7.52	0.24	43.96	26.09	6.08	4.29	0.14
							131.66	4.51	9.01	0.50	0.01
		134.05	16.33	9.08	1.80	0.01	136.30	26.13	9.15	2.86	0.03
							221.61	10.69	11.47	0.93	0.03
		241.61	64.80	11.95	5.42	0.13	249.09	4.44	12.12	0.37	0.03
							270.62	50.31	12.61	3.99	0.13
		441.26	31.85	15.71	2.03	0.09	439.71	26.88	15.69	1.71	0.08
	Stage III						634.60	23.84	18.03	1.32	1.77
							802.73	53.09	19.36	2.74	0.86
		818.05	45.47	19.46	2.34	0.44	825.97	11.16	19.51	0.57	0.13
							910.16	91.49	19.98	4.58	6.73
	Stage II						1434.31	7.62	21.53	0.35	1.07
							1772.82	93.69	21.89	4.28	15.03
		1858.51	24.08	21.94	1.10	0.57	1859.68	15.23	21.94	0.69	0.38
							1925.99	164.47	21.98	7.48	24.50
	Stage I	2498.30	33.27	22.14	1.50	2.43	2488.21	44.17	22.14	2.00	3.74
							2515.02	10.18	22.14	0.46	1.17
							2658.49	22.56	22.16	1.02	14.53
Metamorphic rocks	Stage IV	34.67	11.80	5.73	2.06	0.16	32.96	8.04	5.66	1.42	0.13
		128.92	32.59	8.93	3.65	0.63	127.76	14.92	8.89	1.68	0.36
		236.36	23.84	11.83	2.02	0.20	225.29	8.97	11.56	0.78	0.11
							247.74	6.31	12.09	0.52	0.09
		308.33	20.73	13.40	1.55	0.41	269.04	62.22	12.57	4.95	1.83
		458.56	52.23	15.97	3.27	0.44	435.91	19.27	15.63	1.23	0.16
							495.11	19.05	16.47	1.16	0.19
	Stage III						773.19	41.41	19.17	2.16	0.72
		836.37	110.72	19.57	5.66	1.64	833.07	16.79	19.55	0.86	0.45
							913.31	59.20	20.00	2.96	1.37
	Stage II						1844.16	19.89	21.94	0.91	0.78
		1882.95	89.90	21.96	4.09	2.10	1898.24	72.64	21.97	3.31	2.31
							2164.85	29.78	22.07	1.35	3.59
	Stage I	2507.27	47.37	22.14	2.14	0.96	2507.46	43.49	22.14	1.96	0.90

The statistical results are of great geological significance and correspond well to major geological events in geological history: the Himalayan orogeny, Yanshanian orogeny, Indosinian orogeny, Caledonian orogeny, Jinning orogeny, Lvliang orogeny and Wutai orogeny. Although these results were obtained through big data statistics, which largely eliminated human factors and had strong credibility, there were also some problems affecting the objectivity of the data.

Because of convenience sampling, cluster sampling and hot research sampling, scholars have oversampled zircon in some regions. The existence of these hot data may cause deviations in the big data analytical results. However, hot data are difficult to identify and eliminate. First, geochronological studies around the world are continuing, with the amount of zircon U-Pb data continuing to increase, so there is no objective standard for the quantity of zircon U-Pb age data that a region should contain. Second, when there are too many zircon U-Pb age entries in an area, we do not have a proper and scientific method to determine whether the large quantity in this area is due to hot research or to the original large zircon quantity in this area.

For different lithologies, the sources of hot data ought to be various. Given this situation, the age data of each lithology were separated for statistics, aiming to decipher some implications.

3.2. Single lithology

Similar to the method described in section 3.1, age frequency statistics and Gaussian multipeak fitting were conducted on ²⁰⁶Pb/²³⁸U and ²⁰⁷Pb/²³⁵U ages of each single lithology (sedimentary rock, igneous rock and metamorphic rock), which yielded the mean value, standard deviation, W index, and Y index of zircon growth peaks (Gaussian peaks), as well as the upper limit of the prediction interval for the absolute error of the mean value at the 95% confidence level. The results are shown in Tables 2 and 3 (Figures 2–6).

Figure 2. Gaussian multipeak fitting of zircon growth for great geological events in the whole Earth history: (a) ²⁰⁶Pb/²³⁸U ages of all lithologies; (b) ²⁰⁷Pb/²³⁵U ages of all lithologies; (c) ²⁰⁶Pb/²³⁸U ages of sedimentary rocks; (d) ²⁰⁷Pb/²³⁵U ages of sedimentary rocks; (E) ²⁰⁶Pb/²³⁸U ages of igneous rocks; (f) ²⁰⁷Pb/²³⁵U ages of igneous rocks; (g) ²⁰⁶Pb/²³⁸U ages of metamorphic rocks; (H) ²⁰⁷Pb/²³⁵U ages of metamorphic rocks.

Figure 3. Gaussian multipeak fitting of zircon growth for detailed geological events at each stage using zircons from all lithologies: (a) Stage IV ²⁰⁶Pb/²³⁸U ages; (b) Stage IV ²⁰⁷Pb/²³⁵U ages; (c) Stage III ²⁰⁶Pb/²³⁸U ages; (d) Stage III ²⁰⁷Pb/²³⁵U ages; (E) Stage II ²⁰⁶Pb/²³⁸U ages; (f) Stage II ²⁰⁷Pb/²³⁵U ages; (g) Stage I ²⁰⁶Pb/²³⁸U ages; (H) Stage I ²⁰⁷Pb/²³⁵U ages.

Figure 4. Gaussian multipeak fitting of zircon growth for detailed geological events in each stage using zircons from sedimentary rocks. The plot structure and conventions are the same as those in Figure 3.

Figure 5. Gaussian multipeak fitting of zircon growth for detailed geological events at each stage using zircons from igneous rocks. The plot structure and conventions are the same as those in Figure 3.

Figure 6. Gaussian multipeak fitting of zircon growth for detailed geological events in each stage using zircons from metamorphic rocks. The plot structure and conventions are the same as those in Figure 3.

4. Discussion

4.1. Measurement of hot spot data effect

4.1.1. W index

When certain areas are over-sampled, redundant zircon growth peaks (hot data peaks) would be derived due to these hot data. Fortunately, the statistical method of mean squared weighted deviations or MSWD (Wendt and Carl 1991) provides a reference to analyze the characteristics of hot data peaks. The MSWD is the square of the deviation of each data point from the fitting line divided by the degrees of freedom (the number of data points minus 2), which is commonly used in statistical tests of regression validity, such as the validity of isochronal regression. MSWD measures the degree of data point deviation from the fitting or the degree of data point dispersion relative to the expected dispersion of experimental analytical error. If MSWD < 1, the data points are more concentrated than those expected from experimental analysis errors. From the mathematical point of view, although the data display a good linear relationship, this result actually occurs because the quoted analysis error is overestimated or the error of the experimental instrument is large. If MSWD > 1, the data points are more diffuse than those expected from experimental analysis errors, which indicates that the fitting validity is poor or that it is an error isochrone.

In this study, the idea of MSWD is introduced into the analysis of the W index, which means the ratio of the Gaussian standard deviation to the experimental standard deviation. The premise is that the experimental standard deviation is the upper limit of the prediction interval of the absolute age error fitting curve at the 95% confidence level. If W < 1 or W ≈ 1, the standard deviation of the zircon growth peak is too small, even smaller than the experimental analytical standard deviation. In this case, the peak of zircon growth is probably caused by research focused on regional events, that is, the hot spot effect. Therefore, the Gaussian standard deviation of the hot peak is not more than but roughly at the same level as the experimental analytical error. If W > 1, the span of the zircon growth peak is larger than that of the hot peak. The larger the W, the longer the time span of the zircon growth peak relative to the hot peak is, which can cover a hot peak to a greater extent and weaken the influence of hot spot data on the overall data statistics. Furthermore, as the amount of data increases, the hot data effect decreases.

To analyze the hot spot effect on the zircon growth peaks in geological history, the Gaussian multipeak fitting results of small events in different stages were studied, and their W indexes were analyzed. As described above, the larger the W value is, the smaller the influence of hot data on the peak of zircon growth. Take the statistical results for zircon data of all lithologies as an example. There are some zircon growth peaks corresponding to relatively low W indexes. It is apparent that the Gaussian standard deviations of these peaks are smaller than those of the experiments; thus, the peaks are greatly influenced by hot data, indicating that the peaks with low W indexes should be explained with great caution. The specific zircon growth peaks and their corresponding W indexes are as follows: for ²⁰⁶Pb/²³⁸U ages (Table 2), 130.85 ± 4.25 Ma, 248.16 ± 7.62 Ma, 915.65 ± 15.09 Ma and 1604.68 ± 23.26 Ma correspond to low W indexes of 0.58, 0.86, 0.87 and 0.94, respectively; for ²⁰⁷Pb/²³⁵U ages (Table 3), 131.80 ± 4.72 Ma, 222.71 ± 10.48 Ma and 248.76 ± 6.70 Ma present low W indexes of only 0.52, 0.91 and 0.55, respectively.

Of course, the higher the better is not always true for the W index. A peak of zircon growth with an overly large W index also needs to be constrained. See section 4.1.2 for details.

4.1.2. Y index

Although regional zircon sampling biases lead to hot peaks of zircon growth, this fact does not exclude that scholars carried out geochronological tests on samples of each age in order to study comprehensively, which would produce homogenization effects on age frequency; that is, the age frequency of zircons in each age would tend to be the same. Equation (3) shows that the Gaussian multipeak fitting function has an intercept term y₀, while the Gaussian function actually has no intercept as follows:(6) $f\left(x\right)=a\ast \mathrm{e}\mathrm{x}\mathrm{p}[-(x-\mu {)}^2/2{\sigma }^2]$(6) where a is amplitude, µ is mean and $\sigma$ is standard error. In other words, the intercept term y_0, which does not exist in the real Gaussian distribution, is also a reflection of the homogenization effect. When the homogenization effect is relatively weak, the age frequency of each age does not completely converge but produces a ‘zircon growth peak’ with a very small height but a very large width (i.e. a large time span). To analyze this special hot spot effect, i.e. the homogenization effect, measurement of the Y index is introduced.

The Y index indicates the aspect ratio of a Gaussian peak. The larger the Y index is, the bigger the homogenization effect on the zircon growth peak. If a Y index is too large, i.e. larger than 1, it indicates that the zircon growth peak is too flat to have practical geological significance. This situation can be exemplified by the statistical results for zircon data of all lithologies presented in Tables 2 and 3. The Y indexes of some zircon growth peaks are relatively high. Therefore, these peaks are too flat to have significant geological meanings, even though they have relatively large W indexes and are hardly affected by hot data. In more detail, the zircon growth peaks with ²⁰⁶Pb/²³⁸U ages of 1074.98 ± 78.25 Ma, 1604.68 ± 23.26 Ma, 1729.82 ± 43.85 Ma and 1895.06 ± 87.88 Ma have high Y indexes of 1.76, 1.86, 1.29 and 1.63, respectively (Table 2). In Table 3, the peaks with ²⁰⁷Pb/²³⁵U ages of 1069.97 ± 84.41 Ma, 1595.76 ± 29.08 Ma, 1720.42 ± 36.80 Ma and 1877.04 ± 102.24 Ma, are shown with high Y indexes approximately equal to or larger than 2.

4.2. Time scale of geological events

In regard to the durations of zircon growth events throughout geological history, their variety is apparent. According to the age error fitting curves obtained in section 2.2.1, absolute errors of ²⁰⁶Pb/²³⁸U and ²⁰⁷Pb/²³⁵U ages increase with ages which result in time scales of zircon growth that cannot be classified simply by the standard deviation of a Gaussian peak, since with increasing age, the standard deviation of the Gaussian peak increases as well (shown in Figure 7), displaying a ‘longer’ duration of the geological event. The standard deviations of zircon growth peaks of ²⁰⁶Pb/²³⁸U and ²⁰⁷Pb/²³⁵U ages are drawn in scatter plots. The points on each scatter plot are obviously divided into two regions, i.e. upper left and lower right, as shown in Figure 7. Considering the standard deviation of the zircon growth peak and the age-error, zircon growth events are divided into two scales based on the scatter plots. Given the influence of dating error, zircon growth peaks with younger ages are preferred to distinguish time scales. In detail, standard deviation of 20 Ma (roughly, 10–30Ma) is recommended to be the threshold according to Figure 7.

Figure 7. Distribution of means and standard deviations of zircon growth peaks for all lithologies. (A) ²⁰⁶Pb/²³⁸U ages; (b) ²⁰⁷Pb/²³⁵U ages.

First-order zircon growth peaks (I): the zircon growth during each peak lasts for a long time, i.e. the vertical axis intercept of the classification is larger than 20 Ma. The Gaussian peaks correspond to the Lvliang orogeny, Jinning orogeny, Indosinian orogeny and Yanshanian orogeny. With long time spans, these events were likely caused by plate movements, such as supercontinent breakup, or climate events that last for several cycles, such as snowball Earth events.

Second-order zircon growth peaks (II): the duration of zircon growth is relatively short, i.e. the vertical axis intercept of the classification is smaller than 20 Ma. Such peaks are probably caused by short-term tectonic or climate events, including the Wutai orogeny, Caledonian orogeny, and a series of events around the Lvliang orogeny and Indosinian orogeny. The growth of the earth's crust results from the combination of internal and external dynamic. During the orogenic process, climatic factor could induce crustal decompression and metamorphism, with the growth of zircon. In the past 20 years, researches on young active orogeny, especially Himalayan orogeny, indicate that the dynamics of subduction and mountain-building can be controlled by the processes of erosion, and ultimately climate (Beaumont et al. 2001; Zeitler et al. 2001; Lamb and Davis 2003; Yu et al. 2011; Tu et al. 2015).

For all the long-duration events (i.e. first-order zircon growth peaks), the W indices are high, which means these events are hardly influenced by hot spot data effect. However, events with larger Y indices do not necessarily last long, but with larger difference in statistic results of ²⁰⁶Pb/²³⁸U and ²⁰⁷Pb/²³⁵U ages, maybe for strong homogenization effect.

4.3. Zircon growth of a single lithology

To study the similarities and differences in zircon growth peaks derived from different single lithologies, the growth situations of zircon in section 3.2 were compared with one another, while the Gaussian peaks greatly influenced by hot data were not considered.

Stage I: Sedimentary rocks, igneous rocks and metamorphic rocks all have significant zircon growth peaks close to 2.50 Ga, while igneous rocks also include an active period of zircon growth at approximately 2.65 Ga.

Stage II: Significant zircon growth peaks close to 1.85 Ga are present in sedimentary rocks, igneous rocks and metamorphic rocks.

Stage III: At 0.80–0.82 Ga, there are obvious zircon growth peaks in all three lithologies; it should be noted that before 0.9 Ga, igneous zircons did not grow in large quantities, but there are zircon growth peaks in sedimentary rocks at 0.96 Ga and in metamorphic rocks at 0.93 Ga.

Stage IV: There are apparent zircon growth peaks at ca. 0.44 Ga (sedimentary rock: 437.82 ± 14.85 Ma, igneous rock: 438.46 ± 25.96 Ma, and metamorphic rock: 436.16 ± 19.99 Ma), 0.25–0.27 Ga (sedimentary rock: 249.17 ± 15.95 Ma, igneous rock: 270.98 ± 48.80 Ma, and metamorphic rock: 267.14 ± 59.06 Ma) and ca. 0.13 Ga (sedimentary rock: 133.87 ± 29.90 Ma, igneous rock: 135.40 ± 26.88 Ma, and metamorphic rock: 127.61 ± 14.0 Ma) for all three lithologies. There are also zircon growth peaks at 0.48-0.49 Ga in sedimentary rocks and metamorphic rocks, in detail, 477.30 ± 50.06 Ma in sedimentary rocks and 494.67 ± 15.99 Ma in metamorphic rocks. Moreover, zircon growth peaks are found at 94.33 ± 7.93 Ma, 119.36 ± 9.71 Ma and 300.64 ± 19.04 Ma in sedimentary rocks and at 39.40 ± 23.63 Ma and 220.41 ± 9.50 Ma in igneous rocks.

The growth peaks of zircons from different single lithologies have a generally consistent evolution trend in geological history, which verifies the scientificity of big data analysis as well as the unity between statistical results and geological events. Yet, there are some disagreements, which may result from the formation and evolution characteristics of different types of rocks.

4.4. Zircon growth and geological events

Peak at 2.50 Ga (²⁰⁷Pb/²³⁵U age: 2498.95 ± 49.51 Ma): The peak of zircon growth is consistent with the beginning of the Proterozoic and the period of Wutai orogeny in the North China Craton. The time node of 2.50 Ga, which is the boundary between the Archean and Proterozoic, the end of the era of granite-greenstone terrane development and the onset of an oxygen atmosphere, is of great significance in geological history. During 2.55–2.50 Ga, an abnormally large granitic intrusion was exposed in the eastern part of the North China Craton. A mantle plume and the large amount of heat generated by the plume resulted in widespread melting of the upper mantle and lower crust (Zhao 2009), forming great quantities of granitic magma (TTG), which promoted large-scale formation of zircons. Zircon dating shows that both the Hengshan and Fuping complexes were intruded by major granitoid bodies between 2.52 and 2.48 Ga (Kröner et al. 2005a). Kusky (2011a) proposed that from 2.55–2.5 Ga, an arc terrain (the Wutai arc) with a forearc ophiolite belt on its leading edge (Kusky, Li, and Tucker 2001; Kusky et al. 2004, 2011; Kusky and Li 2010) collided with the passive margin on the western edge of the Eastern Block, above a west-dipping subduction zone (Zhao et al. 2012).

Peak at 1.85 Ga (²⁰⁷Pb/²³⁵U age: 1855.82 ± 107.93 Ma): This zircon growth peak corresponds to the beginning of the Paleoproterozoic Satherian Period and is consistent with the Lvliang orogeny, which occurred at ∼1.85 Ga, representing the main deformation and metamorphism of the Paleoproterozoic basement of the North China Craton. All metamorphic ages obtained for the Trans-North China Orogen are approximately 1.85 Ga, which establishes that the final amalgamation of the Western and Eastern blocks of the craton occurred at ∼1.85 Ga (Zhao et al. 2012). In addition, the Columbia (Nuna) supercontinent was formed at 1.80–1.60 Ga (Pisarevsky et al. 2014; Nordsvan et al. 2018; Kirscher et al. 2019), the time of which is most likely relevant to this peak. At 1.8 Ga, the collision between the northern North China Craton and the Columbia supercontinent (Kusky 2011a) led to a wide range of high-grade metamorphism and obscured previous geological events, containing typical rocks in Mount Wutai and Mount Heng.

Peak at 0.83 Ga (²⁰⁶Pb/²³⁸U age: 828.88 ± 106.77 Ma): This peak corresponds to the Jinning orogeny. The Jinning orogeny represented a collision orogenic event between the Yangtze ancient landmass and the Southeast ancient landmass of China, which occurred from the late Mesoproterozoic to early Neoproterozoic and lasted approximately 200 Ma, from 1000 Ma to 800 Ma (Hao, Li, and Hu 1992). The Jinning orogeny began at the end of Mesoproterozoic (1000 Ma), ended the Mesoproterozoic marine sedimentation in the area and caused the rock to deform and metamorphose. In the early Neoproterozoic (900 ± 50 Ma), magmatic intrusions and large-scale uplift occurred until new transgressive sedimentation began (800 Ma). The area as a whole was on land and experienced weathering and erosion, so that the synorogenic intrusive body was denuded to the Earth's surface. Furthermore, this peak is likely to be relevant to the breakup-convergence cycle of the Rodinia supercontinent, which is a late Mesoproterozoic to Neoproterozoic supercontinent that was assembled at 1100–900 Ma and broke up at 830–720 Ma (Ernst et al. 2008; Ernst et al. 2016; Zhang et al. 2012e; Lü et al. 2018). New observations and age data from China suggest that the breakup of Rodinia started with plume-related magmatism at 830–800 Ma (Zhang et al. 2012e; Cui et al. 2012). It has been suggested that the rifting away of South China from Laurentia was initiated by a ca. 780 Ma mantle plume during the breakup of Rodinia, with the mantle plume activity causing widespread crustal growth in South China in the Neoproterozoic (830–740 Ma) (Wang et al. 2010).

Peak at 0.44 Ga (²⁰⁶Pb/²³⁸U age: 444.29 ± 41.88 Ma): This zircon growth peak corresponds to the boundary between the Ordovician and Silurian Periods during the Caledonian orogeny and Guangxi orogeny (in China). The Guangxi orogeny proposed by Ting (1929) represented a tectonic movement indicated by an unconformity between the Devonian and its underlying strata that prevailed across South China, which began in the Late Ordovician Katian age (453.0 ± 0.7–445.2 ± 1.4 Ma) (Chen et al. 2010) and resulted in large-scale granitic magma emplacement and a small amount of intermediate-basic magmatic activity distributed between the Yangtze and Huaxia regions. The ages of A-type granites and basic rocks formed against an extensional background are concentrated from 450–415 Ma (Yao et al. 2012; Wang et al. 2013; Feng et al. 2014; Zhong et al. 2014), and the time of metamorphism is mainly concentrated between 450 and 420 Ma (Li et al. 2011; Chen et al. 2014; Wang et al. 2013).

Peak at 0.25 Ga (²⁰⁶Pb/²³⁸U age: 249.46 ± 54.07 Ma): This peak corresponds to the boundary between the Paleozoic and Mesozoic Eras in the active stage of Indosinian orogeny. Professor Huang Jiqing proposed the Indosinian cycle in 1945 as the first part of the Alpine cycle in Asia, during which crustal movement of regional folds occurred in the northwest Zhejiang region for the first time (You and Huang 1983). During the Middle and Late Triassic, the main episode of early Indosinian orogeny (Huaiyang orogeny) occurred. In terms of structural changes, the eastern continental margin active belt and other places in China were generally uplifted. Regarding magmatic activity, the magmatic intrusion of the early Indosinian period and the main episode of Indosinian are reflected in the Jiangnan uplift belt and the Fujian-Guangdong, Jiaodong-Liaoning, Jilin-Heilongjiang and Qinling belts (Cui and Li 1983). The Late Permian to Triassic was a critical period in the amalgamation of the South China Craton and is marked by widespread tectonothermal activity, namely the Indosinian orogeny (Yang et al. 2012; Wang et al. 2013).

Peak at 0.13 Ga (²⁰⁶Pb/²³⁸U age: 131.96 ± 18.93 Ma): This peak is consistent with the Yanshanian orogeny. The Yanshanian orogeny was a strong tectonic movement during the Jurassic and Cretaceous in eastern China. Widespread magmatism related to the Yanshanian orogeny has been documented to have formed a very large late Mesozoic magmatic belt in East Asia (Deng et al. 2015). Numerous ages and structural constraints consistently indicate that the Yanshanian orogeny was characterized by alternating stages of crustal shortening at ∼170–136 Ma (Dong et al. 2018) and crustal extension at ∼135–90 Ma (Niu et al. 2004; Li et al. 2013b; Chen et al. 2014; Xia et al. 2016).

Peak at 0.06 Ga (²⁰⁶Pb/²³⁸U age: 58.21 ± 41.06 Ma): This peak corresponds to the Himalayan orogeny. The closure of the northward-subducting Neotethyan lithosphere and the onset of collision between India and Asia to initiate the Himalaya orogeny have been dated to the middle Paleocene (59 ± 1 Ma; DeCelles et al. 2014; Wu et al. 2014; Hu et al. 2015). During the early Himalayan orogeny, the Indian Plate collided widely with Eurasia, and strong tectonic deformation and active tectonic movement occurred in the western part of China, resulting in a widespread unconformity contact relationship. The formation of lithospheric extensional basins in South China and large-scale volcanic eruptions (92–38 Ma) mainly occurred in the Middle and Late Cretaceous to Tertiary. In the early Eocene, the source area of volcanism changed suddenly from the lower crust to the upper mantle, indicating large-scale extension and thinning of the lithosphere in South China at this time (Jia, He, and Lu 2004).

5. Conclusions

We had previously studied the episodes and spatial distribution of zircon growth in the Chinese continental crust, but it is not comprehensive enough (Fang et al. 2018; Wu et al. 2019). In this work, we implemented further statistical analysis and processing on zircon U-Pb ages from the perspective of time.

The single-grain zircon U-Pb chronology database of the Chinese continental crust provides a good platform for big data analysis. By analyzing the chronological data, a series of important zircon growth peak periods in Earth history can be obtained. However, hot data affect the objectivity of the database. To evaluate the impact of hot data on statistical results, the W and Y indexes are introduced in this paper to measure the influences of hot spot and homogenization effects on zircon growth peaks, respectively, by which the range of zircon growth peaks is restricted.

Considering the differences in the time scales of zircon growth peaks, these peaks are divided into two categories. First-order zircon growth peaks (I) have long durations, which may be caused by plate movements or climate events that last for multiple cycles, such as supercontinent breakup and snowball Earth events. Second-order zircon growth peaks (II) have relatively short durations of zircon growth, which may result from factors such as short-term tectonic or climate events. Standard deviation of 20 Ma (roughly, 10–30Ma) is recommended to be the threshold of the vertical axis intercept of the category.

The growth peaks of zircons from different single lithologies are generally similar in geological history. But, there are some disagreements, which may result from the formation and evolution characteristics of different lithologies.

The peak periods of zircon growth obtained in this contribution correspond to major geological events of crustal evolution, which are listed as follows: 2498.95 ± 49.51 Ma, Wutai orogeny; 1855.82 ± 107.93 Ma, Lvliang orogeny; 828.88 ± 106.77 Ma, Jinning orogeny; 444.29 ± 41.88 Ma, Caledonian orogeny; 249.46 ± 54.07 Ma, Indosinian orogeny; 131.96 ± 18.93 Ma, Yanshanian orogeny; and 58.21 ± 41.06 Ma, Himalayan orogeny.

Acknowledgments

We acknowledge the single-grain zircon U-Pb geochronological database of the Chinese continental crust (Fang et al. 2018; Wu et al. 2019), which incorporates data from more than 10,000 articles accessed through the China National Knowledge Infrastructure (CNKI) and Elsevier Science, covering to the end of 2017.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the National Key R&D Program of China, focused special funding [grant number 2017YFC0601203].