Listening to earthworms burrowing and roots growing - acoustic signatures of soil biological activity

We report observations of acoustic emissions (AE) from growing plant roots and burrowing earthworms in soil, as a noninvasive method for monitoring biophysical processes that modify soil structure. AE emanating from earthworm and plants root activity were linked with time-lapse imaging in glass cells. Acoustic waveguides where installed in soil columns to monitor root growth in real time (mimicking field application). The cumulative AE events were in correlation with earthworm burrow lengths and with root growth. The number of AE events recorded from the soil columns with growing maize roots were several orders of magnitude larger than AE emanating from bare soil under similar conditions. The results suggest that AE monitoring may offer a window into largely unobservable dynamics of soil biomechanical processes such as root growth or patterns of earthworm activity - both important soil structure forming processes.

dynamics, but it can exist a delay between the recorded AE and the observed root growth. It is also obvious that part of the root system was not detected through this imaging technique.
After that the maize seeds were added in the glass cell, the roots grew until the 16 th day of the experiment (mean daily root growth rate of 9 cm.day -1 ; Supplementary Table S2). Growth subsequently slowed down (mean daily root growth rate of 0.3 cm.day -1 ). Direct imaging estimated a total of 126 cm of roots along the front of the glass cell. Roots where mostly located in the top 15 cm of the glass cell, around the S2 sensor. This could explain why sensor S2 recorded more events than sensor S3, which was further from the root growth activity.
The AE rates remained low during the first six days of the experiment, i.e. two days before the seeds addition and 4 days after (mean AE rate of 267 and 172 EA.day -1 for sensors S2 and S3; Supplementary Table S2). The AE rates increased drastically over days 6-15 (mean daily AE rate of 2088 and 452 EA.day -1 for sensors S1 and S2). From day 16, the AE rate plateaued until the end of the experiment (mean daily AE rate of 649 and 157 EA.day -1 for sensors S2 and S3). Despite agreement between the AE trends and the observable root growth, the correlation between the daily rates of root growth and AE rates were low when considering the whole experiment (R² of 0.04 and 0.25 for sensors S2 and S3) ( Supplementary Fig. S2). Correlations between AE rates and root growth rate were better when considering the three sub-periods, expected for sensor S2 during days 7-15. This can be due to our limited ability to precisely monitor root growth visually. The root length results from the observable roots in front of the cell. However, more roots were growing into the soil, possibly in contact with the background glass of the cell to which the sensors were connected. This can also explain why we observed a lag between the AE (increasing rate at day six) and the root growth (increasing rate at day two).
To ensure that water movements in the soil (evaporation) did produce negligible AE compared to the root growth, AE were monitored in a glass cell filled with Winzlerboden soil, without seed addition. The conditions were the same as those described in the Materials and Methods section for the plants root growth in glass cell. The soil experienced a steady water evaporation, with a maximum cumulative water evaporation of 4 mm after 2.3 days of experiment ( Supplementary Fig. S3). The monitored AE had a mean daily rate of 127 AE.day -1 , and was lower than the AE rate measured during the first six days of the experiment with addition of maize roots (Supplementary Table S2). This result support the idea that the water movements in soil did produce AE, but in a lower extend that the root growing process.
AE sonification was also conducted using the playitbyr R package 3 (Supplementary Video S3). We detect a lag in Supplementary Video S3 between the observed root growth and the sonified AE due to the roots growing out of view of the camera (in the soil or along the back glass of the glass cell).

Soil structure dynamics and plant root growth in soil columns
During the first two days of the experiment, similar AE rate were registered for upper and lower sensors (S2 and S4 at 5 cm depth and S3 and S5 at 20 cm depth Supplementary Table S3). Mean daily AE rates of 39, 17, 112 and 140 AE.day -1 were recorded for the sensors S2, S4, S3 and S5, respectively. From day three to eight, an increase in AE rates was observed in the column with maize (sensors S4 and S5), in which mean daily AE rates of 2720 and 2701 AE.day -1 were recorded. In contrast, the AE rates in the control bare soil column stayed stable (sensors S2 and S3, mean daily AE rates of 25 and 142 AE.day -1 ).
During the last two days of the experiment, the daily AE rates in the control column (sensors S2 and S3) maintained similar values to those observed previously (24 and 138 AE.day -1 ). The daily AE rates in the column with maize decreased but remained higher than the control column and higher than in the beginning of the experiment (264 and 567 AE.day -1 ).

Results repeatability
The three experiments described in the main text of this paper were re-run in the same conditions as described in the Materials and Methods section to ensure the results repeatability. Supplementary Fig. S4 to S7 show the results of the experiment replicates to assess acoustic emissions from a burrowing earthworm in soil, acoustic emissions from plants root growth in a glass cell, and acoustic emissions from growing plants roots using waveguides in soil columns.
Supplementary Fig. S4 shows the results of the experiment replicate to assess acoustic emissions from the burrowing earthworm in soil. The cumulative length of the earthworm's burrow was 15 cm by the end of the experiment (Supplementary Fig S4f), with 13 cm of tunnels created in the first three days of the experiments. The final two cm were progressively burrowed in the final four days. During this replicate experiment, after a first 13 days, the earthworm re-used the existing tunnels but also continued to create new burrows (Supplementary Fig S4e). The earthworm displacement rate started increasing one day after the beginning of the experiment and continued to increase regularly until the end. This leads us to think that the earthworm activity included both tunnel creation and reuse even in the first three days. Acoustic  Table S4). This result is consistent with the observed earthworm activity that combine tunnel creation and reuse in this phase. During the last four days of the experiments, the AE events were mostly correlated with the earthworm displacement. These results are consistent with the results shown in the main text, illustrating the consistence of the link between AE signature and earthworm activity.
The acoustic emissions generated by the three plants roots growing in the glass cell were not as pronounced in the replication experiment than in the original experiment ( Supplementary Fig. S5).
However, the AE events were still detectable and behave characteristically similar to the visually monitored total root length (Fig. S5a-c). Both measurements come to a plateau towards the end of the experiment. The average AE rates measured during the first six days of the experiments (689 and 375 AE.day -1 for sensors S2 and S3) where higher than those measured during the twelve last days of the experiment (138 and 71 AE.day -1 for sensors S2 and S3) (Supplementary Table S5). This was consistent with the average root growth rate, which was higher during the first six days of the experiments (11 cm.day -1 ) than during the twelve last days of the experiment (2 cm.day -1 ). These results were also consistent with those presented in the main text. The total root growth slowed down and plateaued around day 11 with an observable total root length of about 92 cm. The total root length estimated in the replicate experiment was lower than in the original experiment, which would explain the lower AE events recorded in the replicate. Moreover, despite the more pronounced appearance of the roots in the replication experiment, comparison of the final root distribution along the front and back faces of the glass cell revealed that there were fewer roots in total for the replicate (and no roots were observed on the back face), resulting in significantly reduced AE in the replicate experiment ( Supplementary Fig. S6). The absolute magnitude of the acoustic events detected by the root growth were less than double that of the background noise (thus the normalized events were slightly less than the background noise). The background noise sensor S1 recorded the AE events occurring in the surrounding. It was expose to any occurring disturbance independent to the experiment and its magnitude does not have any significance regarding plant roots growth.
The experimental replicate of the plants root growing in the soil column ( Supplementary Fig. S7) were similar in to the results obtained in the original experiment. The magnitude of the AE signatures were slightly lower by comparison, as well as the AE rates (Supplementary Table S6); however, the acoustic signature was characteristically similar. Nevertheless, only four plants grew during this replicate experiment (compare to the twelve in the experiment exposed in the main text). This could explain the lower AE rates recorded in this experiment. Stem heights reached on average 1.5 cm three days after planting, about 6 cm after five days, about 20 cm after nine days, and about 30 cm at the end of the experiment (after thirteen days). Then, the plants grew up a bit slower than in the original experiment but the stem finally reached longer length. Considering the movement of water in the column (Fig S7b-c), we can see that the large changes in the acoustic signatures occur prior to any rapid changes in the water content. Ultimately, the water movement cannot be the source of the rapid generation of acoustic events in this experiment. These results support the hypothesis that the AE rates can be explained by root growth activity in soil and confirm the potentiality of the method to monitor soil bioturbation processes.
Supplementary Figure S2. Relation between daily AE rate and root growth rate. Solid line for Sensor S2 (y = 1 x 10 -3 x + 5.2). Dashed line for sensor S3 (y = 1 x 10 -2 x + 2.6). The results for sensors S2 and S3 are given after background noise filtering from sensor S1 Supplementary Figure S3. AE monitoring during soil evaporation in the glass cell. a) Cumulative number of acoustic events over time, and b) cumulative water uptake over time. The results for sensor S2 are given after background noise filtering from sensor S1.
Supplementary Figure S4. AE monitoring and earthworm activity in a soil filled glass cell (replicate experiment). Time-lapse images were taken from the front face of the glass cell for the full duration of the experiment (a: beginning and-b: end of the experiment), where X's indicate the locations of the acoustic sensors. The initial packing (a) was augmented by movement of the earthworm (trajectories illustrated in c) culminating in a final perturbed soil state b). (d) Cumulative acoustic events were monitored during the seven days experiment. The results for sensors S2 to S4 are given after background noise filtering from sensor S1. Total cumulative earthworm motion (e) and total length of new tunnels (f) were determined based on the activity monitored using time-lapse images taken from the front face of the glass cell for the full duration of the experiment.
Supplementary Figure S5. AE monitoring during maize roots growing in a soil filled glass cell (replicate experiment). Time-lapse images were taken to monitor maize roots growing in the glass cell from the day the geminated seeds are planted (a) to the last day of the experiment (b), where X's indicate the locations of the acoustic sensors. Cumulative number of acoustic events were monitored for the three separate acoustic sensors (c). The results for sensors S2 and S3 are given after background noise filtering from sensor S1 (the difference between the raw data and the background noise). Simultaneously, the cumulative water uptake was also monitored (d) as well as the estimated total root length (e) determined with timelapse images. The vertical dashed line (c-e) denote the time when germinated seeds were planted in the glass cells (first day of the experiment).