Sedimentary DNA tracks decadal-centennial changes in fish abundance

DNA concentration in core sediments

qPCR analyses for each core (Experiment (a) in Fig. 1) showed that the mean DNA copies for anchovy ranged from 233 ± 215 copies g⁻¹ dry sediment (hereafter, copies g⁻¹, mean ± 1 SD) to 3075 ± 781 copies g⁻¹ (Supplementary Table 1). For all data, anchovy had 1067 ± 968 mean DNA copies g⁻¹. Anchovy DNA copies were detected in all samples, except three (13 cm depth in BG17-1, 41 and 45 cm in BMC18-6). For most samples, DNA was detected in more than three out of four replicates. In each core, the mean DNA copies for sardines ranged from 1.7 ± 4.0 to 12.0 ± 29.2 copies g⁻¹. For all data, sardine had 5.1 ± 3.8 mean DNA copies g⁻¹. This concentration was 0.5% that of anchovy. Sardine DNA copies were not detected in many of the samples from each core. One or two replicates were detected in most samples, while a few samples had more than three out of four replicates for sardines. Jack mackerel DNA copies were also not detected for many samples from each core. Mean DNA copies for each core ranged from 0 to 55.9 ± 127.2 copies g⁻¹. For all data, Jack mackerel had 14.8 ± 19.1 mean DNA copies g⁻¹. This concentration was 1.4% that of anchovy. Few samples had more than three out of four replicates, with most samples having one or two replicates.

For the negative control for sectioning, subsampling, DNA extraction, and PCR processes (Experiment (a) in Fig. 1), DNA of the three marine species was not detected in any of the core sediment samples from Lake Biwa (LBHR18-1) (Supplementary Table 1). DNA was not detected on the PCR blanks either (Supplementary Table 1). Thus, contamination was not an issue during sampling, extraction, purification, and the PCR processes. Also, through the direct sequencing of PCR amplicons by the qPCR assay of Japanese sardine, we confirmed that only the DNA was amplified.

We performed spike test to evaluate the effect of PCR inhibition (Experiment (a) in Fig. 1). All ΔCt values were less than three (ΔCt: −2.4–2.9) (Supplementary Table 2), providing no evidence of inhibition³¹.

Down-core changes in DNA concentration

Down-core changes in DNA concentration for anchovy showed different patterns between each core (Supplementary Figs. 2 and 3) (Experiment (a) in Supplementary Fig. 1). For 50-cm-long core samples, peaks in DNA occurred at around 5 and 20 cm in BMC18-6, but occurred at around 0 cm in BMC17 S1-7 (Supplementary Fig. 2). There was no noticeable peak in BMC17 S1-10 (Supplementary Fig. 2). For 110-cm-long core samples, there was no consistent vertical pattern, except for the uppermost layers, with the highest values being detected for BG18-6W and BG18-8A (Supplementary Fig. 3). For sardine and jack mackerel, there were no consistent vertical patterns in DNA for the short cores. In contrast, the 1.1-m-long cores had peaks centered at around 16 and 57 cm deep for sardine and at around 20 cm for jack mackerel. Comparison between short and long cores (Supplementary Figs. 2 and 3) showed that anchovy and jack mackerel had the highest DNA concentrations in the uppermost layers in the short cores. The long cores did not show a similar trend, due to loss of surface layers (approximately 20 cm) during core collection. In contrast, the highest DNA concentrations for sardine occurred at 57–58 cm depth in the long cores, not in the uppermost layers of the short cores.

There was no clear evidence that DNA concentrations were higher in core samples that were instantly frozen after core collection (core BMC17 S1-7) compared to samples that were frozen 6 days or 4 weeks after core collection (Supplementary Figs. 2 and 3) (Experiment (a) in Fig. 1). Thus, chilled storage for 4 weeks only caused minor degradation of DNA in core samples.

Temporal changes in DNA concentration

General additive models (GAMs) showed that the decadal–centennial dynamics of the inter-core, seven-year averaged, and sedDNA concentrations for the last 300 years significantly varied non-linearly (Japanese anchovy, s = 7.22, P = 2.96 × 10⁻⁷; Japanese sardine, s = 12.61, P = 1.10 × 10⁻⁴; jack mackerel, s = 8.831, P = 2.84 × 10⁻⁹, Fig. 2). DNA concentrations for Japanese anchovy were high after 2010 CE (BMC18-6, BMC17 S1-7, and BMC17 S1-10) (Fig. 3). While there was no consistent pattern in the time series of the cores before 2000 CE, one or two of the time series showed high values around the 1960s CE and the 2000 CE. These periods with high values showed large scatters between the cores, indicating spatial heterogeneity in DNA deposition.

These high values were not obtained in the anchovy fish scale concentrations during the same periods¹⁷ (Supplementary Fig. 4). There was also no significant relationship in the Type II regression model for the inter-core, seven-year averaged, concentrations in DNA (Fig. 4a) with those of fish scale concentrations (Fig. 4c) for the 300 years (R² = 0.0157, P = 0.429, n = 42, Fig. 5a, also see Supplementary Fig. 5 for the log-transformed model). In contrast, the two DNA peaks in the 1960s and 2000 were temporally consistent with those of the catch record in Japan (Statistics of Agriculture, Forestry and Fisheries, with the landing data being updated from previous studies^32,33) (Fig. 4b). This result was supported by a Type II regression model, with a significant correlation existing between inter-core, seven-year averaged concentrations in DNA (Fig. 4a) and seven-year averaged catches in Japan (Fig. 4b) (R² = 0.255, P = 0.0459, n = 16, Fig. 6a, also see Supplementary Fig. 6a for the log-transformed model). Anchovy sedDNA and landings in Japan before 1990 showed a positive-phase relationship with the Bungo Channel (Supplementary Fig. 7, see Supplementary Fig. 1a for the location), but a negative-phase relationship with the central Seto Inland Sea (Supplementary Fig. 7c). A decadal peak around 2000, as shown by the sedDNA and landings in Japan, was not obtained in the landings from the Bungo Channel and Beppu Bay, Iyo-nada, and Suo-nada (Supplementary Fig 7a, b, see Supplementary Fig. 1 for the locations and see Supplementary Discussion for the reasons). An abnormally high value in 2014–2017 was not found in the landing records (Supplementary Fig 7a, b). This inconsistency indicates the presence of enriched DNA in the surface layer that is susceptible to rapid decomposition due to early diagenesis in a few years.

DNA concentrations for Japanese sardine showed large scatters of contemporary values between the cores (Fig. 3b), indicating spatial heterogeneity in DNA depositions. The concentrations were high (>20 copies g⁻¹) during the 1840s to 1850s and 1970s to 1980s for the time series of the three cores, and were low (<20 copies g⁻¹) during the other periods (Fig. 3b). GAMs showed that the peaks during the 1840s to 1850s were comparable to, or higher than, those during the 1970s to 1980s (Fig. 2). These two peaks were consistent in time with the high values (>0.25 g⁻¹) recorded for sardine fish scale concentrations (Supplementary Fig. 4b, f); however, there was no noticeable peak during the 1920s and 1930s (Figs. 3b and 4b), despite a peak occurring in the fish scale record (Fig. 4f and Supplementary Fig. 4b). The peak in DNA during the1970s to 1980s corresponded to a distinct peak in sardine catches during the twentieth century (Fig. 4e). The DNA peak in the 1840s to 1850s was consistent with good catch periods recorded by historical documents in and around the Bungo Channel^34,35 (Fig. 3b). Sardine DNA was detected during ~1700 CE (14 copies g⁻¹), which was consistent with a good catch period recorded in the historical documents (Fig. 3b), and a minor peak in the fish scale record (Fig. 4f and Supplementary Fig. 4b). Type II regression for sardine showed a significant correlation between inter-core, seven-year averaged concentrations of DNA (Fig. 4d) and fish scale concentrations (Fig. 4f) for the last 300 years (R² = 0.436, P = 1.93 × 10⁻⁶, n = 42, Fig. 5b, also see Supplementary Fig. 5b for the log-transformed model). It also showed a significant correlation between inter-core, seven-year averaged concentrations of DNA and seven-year averaged catches in Japan (R² = 0.269, P = 0.0395, n = 16, Fig. 6b, also see Supplementary Fig. 6b for the log-transformed model). sedDNA and landings in Japan showed a positive-phase relationship with Bungo Channel. However, a clear relationship was not detected with the landings in Beppu Bay, Iyo-nada, and Suo-nada or the central Seto Inland Sea (Supplementary Fig. 8, see Supplementary Discussion for the reasons). There was a negative phase relationship between sardine and anchovy in sedDNA after the 1950 CE (Fig. 2a, b).

Jack mackerel DNA concentrations for each core (Fig. 3c) were high, exceeding 50 copies g⁻¹ around 1970 and 1990 for the two core time series, and exceeding 100 copies g⁻¹ after 2005, with low values (<20 copies g⁻¹) being obtained during the other periods. Before 1965, DNA values were low, with less than 50 copies g⁻¹. There was no significant correlation between inter-core, seven-year averaged concentrations of jack mackerel DNA (Fig. 4g) and landings in Japan (Fig. 4h) (R² = 0.0786, P = 0.501, n = 8, Fig. 6c, also see Supplementary Fig. 6 for the log-transformed model). This non-correlation might be due to the small sample size (short record of jack mackerel landing). Therefore, we further compared temporal variation in jack mackerel DNA with longer-available time series of Caranginae in Japan, which did not discriminate between landings of Jack mackerel and Amberstripe scad before 1960. However, variation in the annual catch of Caranginae might reflect that of jack mackerel, because jack mackerel catches after 1960 accounted for 79 ± 14% of Caranginae landings (sum of jack mackerel and amberstripe scad). There was a significant correlation between inter-core, 7-year averaged concentrations of jack mackerel DNA (Fig. 4g) and 7-year averaged catches of Caranginae in Japan (Fig. 4h) (R² = 0.453, P = 0.00221, n = 18, Fig. 6d, also see Supplementary Fig. 6d for the log-transformed model). Jack mackerel sedDNA and landings in Japan showed a positive-phase relationship with landings from the Bungo Channel, and a weak positive-phase relationship with landings from Beppu Bay, Iyo-nada, and Suo-nada (Supplementary Fig. 9). The 2017 peak was not found in the landing records (Supplementary Fig. 9). This inconsistency indicates the presence of enriched DNA in the surface layer that is susceptible to rapid decomposition due to early diagenesis in a few years.

Relationship between anchovy DNA and environmental factors

Anchovy DNA copies showed a weak significant negative correlation with titanium (Ti) contents (a soil mineral tracer³⁶) (r = −0.41, P = 0.011), and no significant correlation with total organic carbon contents (TOC, a mixture of terrestrial and marine-derived organic carbon), molar carbon and nitrogen ratios (C/N, an index of supply rate of terrestrial organic matters³⁷), biogenic opal (mainly originated from diatom valves), and sedimentation rates (Supplementary Table 3). In the upper laminated layers (Supplementary Figs. 11 and 12), anchovy DNA showed no correlation with any of the environmental indices (r = −0.36–0.08, P > 0.05, Supplementary Table 3). In the lower massive layers (Supplementary Figs. 11 and 12), anchovy DNA showed a significant positive correlation with TOC (r = −0.50, P = 0.029) and biogenic opal (r = 0.47, P = 0.044), and a negative correlation with C/N (r = −0.48, P = 0.040). It showed no correlation with Ti and sedimentation rate.

Source materials of DNA in marine sediments

The DNA in the pore water of each sample (Experiment (c) in Fig. 1) was not detected by qPCR assays for any of the species (Table 1). In contrast, anchovy DNA was detected in the residual bulk sediments of all samples (range: 367–6423 copies g⁻¹, mean: 2704 ± 2233 copies g⁻¹), while sardine DNA was detected in two samples (range: 12.4–283.5 copies g⁻¹, mean: 51.6 ± 113 copies g⁻¹) and jack mackerel was detected in one sample (84.6 copies g⁻¹) (Table 1, Experiment (c) in Fig. 1). DNA was only detected in the fish scales of anchovy (1.5 ± 4.2 copies scale⁻¹) (Table 2) (Experiment (b) in Fig. 1). DNA from bones was not detected in any of the species (Table 2) (Experiment (b) in Fig. 1). DNA was detected in the 63–180 μm size fractions of one sample for sardine (0.9 ± 1.7 copies g⁻¹ dry sediment before sieved) and jack mackerel (0.3 ± 0.6 copies g⁻¹ dry sediment before sieved), but was not detected for anchovy (Table 2). DNA was not detected in the 180–500 μm fractions for any of the species (Table 2).

Source: Ecology - nature.com