6.1 Changing Provenance
Before using the changing zircon U-Pb age spectra to infer changing sediment provenance we examine the possible role of grain size in controlling the results. Sediment grains are fractionated during transport because different densities and shapes affect their settling characteristics [Garzanti et al. , 2009]. Zircons all have the same density but the size and shape of the grains from a given source may vary and influence the final conclusions. If one source is associated with smaller or larger grains compared to other sources then this may prejudice the analysis, especially if the grains are too small to be analyzed. Garzanti et al. [2009] concluded that this effect was moderate in the Ganges-Brahmaputra catchment, which has strong similarities to the Indus. We plot major, provenance-related age populations (0–25, 40–70, 70–120, 300–750, 750–1250 and 1500–2300 Ma) against median grain size for all samples considered here to see if grain size plays a strong role in controlling the age spectra. Figure 10 shows that there is not a strong correlation between sediment median grain size and the proportion of various provenance-sensitive age groups. However, we note that the four coarsest sediments (>100 µm) do contain more 750–1250 and 1500–2300 Ma grains compared to the 40–70 and 70–120 Ma groups. The effect is especially strong with the 1500–2300 Ma group. In contrast to work on the Amazon River by Lawrence et al. [2011] who showed that older grains were significantly smaller than younger ones, the reverse may be true in the Indus. It is however noteworthy that the coarser sediments are all young 3.02 Ma and younger and as demonstrated below the provenance inferred from similar aged finer sediment is not greatly different and also consistent with neighboring bulk sediment Nd isotope constraints. We conclude that there may be a grain size issue with the coarsest sediment, but that this is not dominant in controlling the U-Pb age spectra.
The zircon U-Pb age spectra are used to track the source evolution of sediment reaching the Arabian Sea and compared to bedrock zircon U-Pb age signatures of possible source areas (Fig. 7). The abundance of grains younger than 200 Ma correlates well with young bedrock from the Indus Suture Zone, particularly in Kohistan, the Transhimalaya and Karakoram, as well as Nanga Parbat (Fig. 8). The abundance of these young zircon grains clearly points to sediment being supplied by the Indus River and not by peninsular India, where no magmatism <200 Ma is known outside the Deccan Plateau. Detrital zircon grains older than 350 Ma also largely correlate with various bedrock sources known in the Himalaya. Detrital zircon age modes between 350 and 750 Ma have been correlated with bedrock sources in the Tethyan Himalaya [Alizai et al. , 2011], although it is generally agreed that there is little real difference in terms of U-Pb ages, between Tethyan and Greater Himalaya zircon signatures [Gehrels et al. , 2011], and these are in any case not always mapped consistently by different groups [Webb , 2013]. Consequently, zircons with ages between 350 and 1250 Ma could be derived from either source. The older samples show relatively low abundance of grains in this age range, but these increased significantly starting at 5.87 Ma and become very abundant in the last few million years. Older grains, dating between 1500 and 2300 Ma, are particularly common in Lesser Himalayan sources, although they are also present in smaller amounts in the Tethyan and Greater Himalaya [DeCelles et al. , 2000; Gehrels et al. , 2011]. These mainly Paleoproterozoic zircon grains are almost entirely absent from the Laxmi Basin Miocene samples, but show a marked increase beginning at 5.72 Ma, and becoming very abundant beginning at 1.56 Ma (Fig. 7). We therefore interpret these patterns to indicate a progressive increase in erosion from the Himalaya starting after 7.0 Ma, and especially starting at 5.72 Ma, with strong erosion from the Tethyan and Greater Himalaya. After 3.02 Ma there is a dramatic increase in erosional flux from the Lesser Himalaya, which have had a strong influence on the river system since the onset of the Holocene [Clift et al. , 2004; Clift et al. , 2008a].
If we only consider the zircon grains younger than 200 Ma then we can see that there is evidence of erosion, from both Kohistan and from the Karakoram, in most of the samples analyzed (Fig. 8). Kohistan is particularly noteworthy for having zircon dated between 40 and 70 Ma [Alizai et al. , 2011; Zhuang et al. , 2018], although there are similar aged units in the Karakoram as well. However, zircon grains older than 70 Ma but younger than 120 Ma are almost exclusively known only from Karakoram bedrock sources [Searle , 1996]. The 3.17 Ma sample does not show the younger 40–70 Ma population, suggesting that it did not receive any significant material from Kohistan/Ladakh.
The youngest (<25 Ma) zircon grains are more enigmatic in terms of their provenance. While very young zircons are known from the present-day Nanga Parbat massif, these are generally younger even than the 25 Ma zircon U-Pb age component observed in many of the samples [Zeitler et al. , 1993]. Our new data also show an increased influx from bedrock sources with very young zircon starting at 3.02 Ma, as well as a brief appearance at around 5.78 Ma. It is possible that this increase starting at 3.02 Ma reflects the emergence of Nanga Parbat, although we cannot exclude the influence of other young sources in the southern Karakoram metamorphic belt, which also contains rocks of this age and have experienced very rapid exhumation in the last few million years [Wallis et al. , 2016]. Because the Deccan Plateau volcanic rocks were erupted rather quickly around 65 Ma, it is hard to completely exclude their influence because grains of a similar age are also known in Kohistan, and in the Karakoram. However, the erosion from the Deccan Plateau would not account for the other young grains and an influx from that area should result in a clear peak age at 65 Ma, which is not observed.
We also assessed the evolving provenance patterns of sediments in Laxmi Basin using a multidimensional scalar (MDS) analysis of the detrital zircon U-Pb dates [Vermeesch et al. , 2016]. In this plot, which is a type of principle component analysis, samples with similar age spectra plot close to one another, while distinct samples are far separated. Figure 11A shows all the detrital samples data, along with a modern river mouth and a delta sample (KB-40) dating from shortly after the LGM [Clift et al. , 2008a]. The MDS analysis shows clear and coherent patterns. Samples deposited at and after 1.56 Ma, are relatively similar to the modern river. In contrast, the oldest samples plot in a cluster suggesting a similar Miocene provenance and a subsequent progressive shift from right to left with decreasing depositional age, although with some reversals, most notably at 3.17 and 3.57 Ma. This reflects an overall shift in the zircon age spectra through time. Nonetheless, the LGM sample has stronger similarities with sediments deposited on the fan during the Late Miocene. Earlier work implied that erosion during the LGM was focused in the Karakoram [Clift et al. , 2008a] compared to the modern river or during the Holocene when the summer monsoon was strong [Caley et al. , 2014; Fleitmann et al. , 2003; Gupta et al. , 2003]. The new data indicate that older Miocene samples were also deriving their material from Karakoram sources, and this was followed by a shift to more Himalayan sources, especially in the last few million years. The plot implies that the change might be step wise, with a change starting between 7.0 and 5.87 Ma and again at 1.56 Ma.
The fact that the youngest turbidite sands are most similar to the modern interglacial river, and not the compositions of the Indus shortly after the LGM, also implies that most of the sediment deposited in the Indus Fan has been eroded during interglacial times when the monsoon was strong, even if final deposition did not occur until the sea level fell during the onset of the subsequent glaciation. We envisage fast interglacial erosion generating great volumes of sediment, which is then mobilized, transported, and delivered to the delta as the rains strengthened [Jonell et al. , 2017]. The sediment would then be stored on the shelf or in the upper canyon during sea level high stands before being eroded and redeposited as sea level fell [Li et al. , 2018]. This emphasizes the importance of monsoon intensity in controlling erosion and sediment delivery in the Western Himalaya.
We also compared the Arabian Sea sediments with known zircon ages from bedrock sources themselves. Figure 11B shows the progressive changes from the Miocene to the present and emphasizes the fact that the stratigraphically oldest detrital zircon samples plot closest to sources in the Karakoram and have similarities with analyses from the trunk stream (upper reaches) of the main Indus River, before it mixes with the Himalaya-draining Eastern tributaries, such as the Jhellum, Chenab, Ravi, Sutlej and Beas (Fig. 1). Conversely, the stratigraphically youngest sediments plot on this diagram closest to Himalayan sources and have greater similarity not only to the modern river mouth, but also Himalayan tributaries such as the Ravi, Chenab and Jhellum rivers.
These data also imply that Nanga Parbat has not been a very important contributor to the bulk sediment flux. Whether this is actually true or not is not entirely apparent because the bedrock analyses from Nanga Parbat were focused on igneous rocks in the center of that metamorphic massif, and might not be representative of the net erosional flux from this particular source. However, the relationships displayed in Figure 11B can be readily explained as a simple mixing between Karakoram and Himalayan sources, with a progressive shift towards the Himalaya through time.