Tuned in: plant roots use sound to locate water

Abstract

Because water is essential to life, organisms have evolved a wide range of strategies to cope with water limitations, including actively searching for their preferred moisture levels to avoid dehydration. Plants use moisture gradients to direct their roots through the soil once a water source is detected, but how they first detect the source is unknown. We used the model plant Pisum sativum to investigate the mechanism by which roots sense and locate water. We found that roots were able to locate a water source by sensing the vibrations generated by water moving inside pipes, even in the absence of substrate moisture. When both moisture and acoustic cues were available, roots preferentially used moisture in the soil over acoustic vibrations, suggesting that acoustic gradients enable roots to broadly detect a water source at a distance, while moisture gradients help them to reach their target more accurately. Our results also showed that the presence of noise affected the abilities of roots to perceive and respond correctly to the surrounding soundscape. These findings highlight the urgent need to better understand the ecological role of sound and the consequences of acoustic pollution for plant as well as animal populations.

Full text

1 3
Oecologia
DOI 10.1007/s00442-017-3862-z
BEHAVIORAL ECOLOGY –ORIGINAL RESEARCH
Tuned in: plant roots use sound to locate water
Monica Gagliano1 · Mavra Grimonprez1 · Martial Depczynski2,3 · Michael Renton4
Received: 31 October 2016 / Accepted: 31 March 2017
© Springer-Verlag Berlin Heidelberg 2017
the abilities of roots to perceive and respond correctly to
the surrounding soundscape. These findings highlight the
urgent need to better understand the ecological role of
sound and the consequences of acoustic pollution for plant
as well as animal populations.
Keywords Foraging behavior · Hydrotropism · Moisture
sensing · Bioacoustics · Directional root growth
Introduction
All living organisms have basic needs and can only sur-
vive in environments where vital resources are available
for those needs to be met. Water is one of those essential
resources and its availability plays a critical role in terres-
trial ecosystems where it strongly influences abundance,
spatial distribution and species interactions of a wide range
of plant and animal groups (Hawkins et al. 2003; McCluney
and Sabo 2009; McCluney et al. 2012; Ledger et al. 2013).
Because water is often limited and can be unevenly distrib-
uted across time and space, both animals and plants have
evolved a number of morphological and physiological traits
as well as behavioral strategies to cope with water scarcity
and avoid dehydration. Ultimately when faced with water
scarcity, both animals and plants have two main options:
water-saving or water-seeking.
Several animals and plants have evolved to cope with
water scarcity through their impressive physiological
capacity to save previously acquired water (e.g., cam-
els, Bekele et al. 2013; cacti, Niklas 1997). Taken to an
extreme, bryophytes like the so-called ‘resurrection plants’
can remain in a dried state for years and then, rehydrate
and return to a fully functional state within 48 h of rain
(Scott 2000). These and many other morphological and
Abstract Because water is essential to life, organisms
have evolved a wide range of strategies to cope with water
limitations, including actively searching for their preferred
moisture levels to avoid dehydration. Plants use moisture
gradients to direct their roots through the soil once a water
source is detected, but how they first detect the source is
unknown. We used the model plant Pisum sativum to inves-
tigate the mechanism by which roots sense and locate
water. We found that roots were able to locate a water
source by sensing the vibrations generated by water mov-
ing inside pipes, even in the absence of substrate moisture.
When both moisture and acoustic cues were available, roots
preferentially used moisture in the soil over acoustic vibra-
tions, suggesting that acoustic gradients enable roots to
broadly detect a water source at a distance, while moisture
gradients help them to reach their target more accurately.
Our results also showed that the presence of noise affected
Communicated by Hermann Heilmeier.
Electronic supplementary material The online version of this
article (doi:10.1007/s00442-017-3862-z) contains supplementary
material, which is available to authorized users.
* Monica Gagliano
monica.gagliano@uwa.edu.au
1 Centre for Evolutionary Biology, School of Animal Biology,
University of Western Australia, Crawley, WA 6009,
Australia
2 Australian Institute of Marine Science, Crawley, WA 6009,
Australia
3 Oceans Institute, University of Western Australia, Crawley,
WA, Australia
4 School of Plant Biology, University of Western Australia,
Crawley, WA 6009, Australia

Oecologia
1 3
physiological adaptations are clearly useful water-saving
or desiccation-tolerance mechanisms; in general, however,
the most common strategy to economize on water involves
changes in behavior. By reducing rates of mobility during
the hottest hours of the day, animals are able to minimize
water loss. In plants, changes in the orientation of leaves
(i.e., paraheliotropism) to avoid light and reduce leaf tem-
perature can also substantially increase water use efficiency
(Bielenberg et al. 2003).
Despite the incredible water-saving abilities displayed
by many species, the majority of animals and plants are
not physiologically well-equipped to survive long periods
of drought without searching for new water sources. Con-
sequently, they have evolved water-seeking behaviors to
deal with and actively search for more suitable (wetter)
environments (animals, McCluney and Sabo 2009; plants,
Kiss 2007; Cassab et al. 2013). To actively search and find
water, both plants and animals must rely on information of
various kinds to make the most efficient directional deci-
sions. Animals are known to use an array of multisensory
orientation systems, which may include visual, auditory,
olfactory, magnetic, hygrotactic, anemotactic, polarotactic
and other cues (Bernáth et al. 2004; Russell et al. 2014).
Plants are also known to be exquisitely sensitive to a wide
range of environmental cues including geomagnetic fields
and moisture gradients, which they use to direct their roots
through the soil once a water source is detected (Hart
1990). However, how plants sense and are able to move in
the direction of water in the absence of a moisture gradient
still need to be elucidated.
Investigations addressing this fundamental question on
the water-seeking abilities of roots have been extremely
limited (Cassab et al. 2013). The main reason for this
paucity is that hydrotropism is considered a “weak”
tropism relative to other tropisms such as phototropism
and gravitropism (for example, experiments have been
performed in space in order to mitigate the overwhelm-
ing effects of gravity and gravitropism; Wolverton and
Kiss 2009; Kiss et al. 2012). Nevertheless, this paucity is
perplexing given that this behavior is critical to the way
plants acquire water and hence survive. Here, we tack-
led this issue by examining the water-finding ability of
roots of pea seedlings (Pisum sativum), a model species
that has previously been used for assessing root hydro-
tropic response to moisture gradients (Jaffe et al. 1985).
Recent work has found that the roots of corn seedlings
are able to detect sound waves or vibrations and selec-
tively use them for orientation (Gagliano et al. 2012a). As
sound travels readily and far in dense environments such
as soil, a plant’s ability to detect vibrations may represent
a very efficient, yet hitherto unexplored, way of captur-
ing information from distant sound sources for orienta-
tion towards water. For example, roots may detect noise
emanating from water moving through the soil or flowing
through natural channels or human-made structures such
as underground pipelines used in water supply networks
and sewer systems. As a matter of fact, the invasion of
sewer pipes by tree roots is an all too common and costly
issue in municipalities around the world (United States
Environmental Protection Agency 1999; Östberg et al.
2012; Xie et al. 2014), yet no research has been directed
towards better understanding plant hydrotropic behavior
in the context of bioacoustics. Whether acoustic cues are
really contributing to root orientation towards a water
source available in the soil remains to be seen, however.
Therefore, in this study, we used a custom-designed
Y-maze (cf. Fig. 1) to investigate how roots locate a water
source, by experimentally testing how the behavior of
roots responds to different acoustic cues. The study con-
sisted of a series of test scenarios (TS) where roots effec-
tively ‘chose’ between different treatments applied to
the ends of the Y-maze. Specifically, we experimentally
investigated how roots choose the direction that correctly
leads them to water by testing their response to the sound
of water moving inside a pipe (Experiment 1) and then
using playback experiments to test whether roots respond
to sound recordings of water (Experiment 2). We also
used recordings to determine whether roots were able to
discriminate between water and other sounds when these
co-occur (Experiment 3).
Fig. 1 Schematic representation of the custom-designed experimen-
tal Y-maze, made of a PVC pipe filled with soil and attached to two
tightly fitting small black plastic pots and two transparent rectangular
plastic trays at each lower end. Not to scale

Oecologia
1 3
Materials and methods
Germination, growth conditions and Y‑maze design
Seedlings of the garden pea (Pisum sativum cv Mas-
sey Gem) to be tested in the Y-maze trials were ger-
minated hydroponically in 250 mL round containers.
Seeds were firstly soaked in water for 24 h and then
wrapped with clean wet paper-towel and an exter-
nal layer of aluminum foil. Five seeds per roll were
used and seed rolls were placed vertically in a round
container, immersed in 50 mL of water (replenished
daily) and incubated in a dark germination chamber at
24 °C ± 0.1 (SE) average temperature and 62% ± 0.3
(SE) average humidity (simultaneously recorded using
a HOBO data-logger). Progress towards germination
was evaluated each morning and seeds were considered
to have germinated when the radicle was >5 mm long.
Upon germination, each seedling was planted in the
center of its individual custom-designed Y-maze (Fig. 1)
at a depth of approximately 25 mm, after the maze had
been filled with soil (Osmocote seed raising and cut-
ting mix) and the soil had been saturated with water and
then allowed to drain freely for about 5–10 min (after
which water stopped draining). Each maze consisted of
an inverted Y-shaped PVC pipe fitted with two tightly
fitting small black plastic pots (55 mm diameter by
47 mm in depth) and two transparent rectangular plastic
trays (90 mm × 70 mm × 40 mm) at each lower end.
Each maze was secured to a polyurethane foam base
and placed into a plastic planting tray. Each seeded
maze was randomly allocated to a test scenario (details
below). To ensure similar growth conditions across test
scenarios, seeded mazes were haphazardly distributed
in a small (2 m × 4 m) glasshouse at the University of
Western Australia Botany Glasshouse complex, and then
left to grow undisturbed under natural light conditions
for 5 days. For the entire duration of the experiments,
the glasshouse was temperature-regulated by ventila-
tion fans and automated shade-screening systems and
its environment was monitored and recorded at 15-min
intervals by a sensor located in the center of the room
[average temperature 23 °C ± 0.1 (SE); average relative
humidity 50% ± 0.3 (SE)]. In addition, air temperature
was measured three times daily by two probes, ran-
domly positioned adjacent to the seeded mazes inside
the glasshouse [average temperature: 24 °C ± 0.2 (SE)].
Sunlight in the glasshouse was recorded every 30 min
by a roof-mounted sensor [average ambient light level:
356 μmol m−2 s−1 ± 5.7 (SE)]. A total of 89 seeded
mazes were included in the study.
Experiment 1: root directional responses to the sound
of water inside a pipe
The aim of this experiment was to investigate whether
acoustic cues, and specifically the sound of moving water,
contribute to the hydrotropic behavior of roots. In test sce-
nario 1 (TS1) (see Table 1, n = 10; Fig. 2a), we established
the extent of root directional growth towards an actual
water gradient produced by the presence of 100 ml of water
contained in the transparent plastic tray at the base of one
side of the maze [WTR]. In this baseline scenario, the water
source was located on one side of the maze only (A) versus
nothing applied to the other side (B). The A side was desig-
nated as the treatment side and assigned to the left or right
side at random for each replicate maze. On day 5, the small
black pots at the base of the maze were removed to expose
the position of the primary root in the maze [i.e., left (L)
versus right (R) side, (A) vs (B) treatment; Fig. 3a] and the
test terminated. At the end of each test, the soil was gen-
tly washed out of the maze to reveal the distribution of the
root system (primary root and lateral roots) within the maze
while preventing damage. The position of the primary root
was recorded and seedlings were then carefully extracted
from the maze and photographed against it (Fig. 3b).
In test scenario 2 (TS2) we examined the directional
behavior of roots in response to one discrete test scenario
in which seedlings had no direct access to water (Table 1,
n = 10; Fig. 2b). In this scenario, the treatment consisted of
the live sound of water running through clear PVC flexible
tubing wrapped around one of the two plastic pots at the
base of each maze with water continuously re-circulated
by an SP-980 aquarium pump [WTR-PIPE]. We compared
the observed frequency of seedlings directing their primary
root towards the sound of water in this scenario relative
to the expected frequency represented by seedlings in the
baseline WTR group from TS1.
An additional separate experiment was also conducted
to establish whether the presence of circulating water in the
flexible tubing attached to one side of the maze might have
had an effect on the soil temperature within the maze and
thus root orientation (see details in Electronic Supplemen-
tary Material).
Experiment 2: root responses to recorded acoustic cues
The aim here was to establish whether roots selectively
respond to the sound of water. In test scenario 3 (TS3;
Table 1, n = 10) we used playback experiments to first test
the directional behavior of roots responding to the recorded
sound of water running through a pipe [(WTR-REC); see
Fig. S1 for details]. In test scenario 4 (TS4), we examined

Oecologia
1 3
the directional growth behavior of roots in response to a
test scenario, where the playback recording treatment was
computer-generated white noise [NOISE; n = 10; Table 1].
As above, both test scenarios involved the acoustic treat-
ment being applied to one side of the maze (A) versus
nothing applied to the other side (B), with A assigned to
left or right side at random for each maze. In both these test
scenarios, seedlings had no direct access to water.
To account for the possible artifact effect caused by the
presence of the sound equipment itself, we also tested root
responses to two additional control scenarios (TS5, TS6),
where either the sound equipment was turned on and broad-
casted recorded silence [ZERO Hz; n = 10] or the sound
equipment was turned on but not playing [NOT PLAYING;
n = 10] (see Table 1; Fig. 2).
All recorded sound treatments were played on continu-
ous loops throughout night and day, using portable MP3
players and small vibration speakers attached directly to the
black plastic pot at the base of the maze on the randomly
Table 1 Treatments applied in each test scenario, together with the expected direction of each potential hypothesized effect in each scenario: acting towards (1) or against (−1) root growth in
the A treatment direction, or having no influence (0)
When an effect is acting on both sides equally, such as equipment in TS8 and TS9, it is assumed to have no net influence towards or against root growth in the A treatment direction (0)
Exp TS Treatment A Treatment B Hypothesized effects
Water contact Water presence Sound Water sound White noise Equipment presence Equipment on
1 TS1 Water (WTR) Nothing 1 1 0 0 0 0 0
TS2 Sound of water through pipe (WTR-PIPE) Nothing 0 1 1 1 0 0 0
2 TS3 Recorded sound of water (WTR-REC) Nothing 0 0 1 1 0 1 1
TS4 White noise (NOISE) Nothing 0 0 1 0 1 1 1
TS5 Zero Hz (ZERO HZ) Nothing 0 0 0 0 0 1 1
TS6 Sound equipment on but not playing (NOT
PLAYING)
Nothing 0 0 0 0 0 1 0
3TS7 Recorded sound of water (WTR-REC) Water (WTR) −1−1 1 1 0 1 1
TS8 Recorded sound of water (WTR-REC) White noise (NOISE) 0 0 0 1 −1 0 0
TS9 Recorded sound of water (WTR-REC) Zero Hz (ZERO HZ) 0 0 1 1 0 0 0
Fig. 2 Schematic representation of experimental treatments, where a
water was directly accessible (WTR) or b present inside the tubing
but not accessible (WTR-PIPE), and where c the recorded sound of
water (WTR-REC), computer-generated white noise (NOISE) or no
sound (ZERO Hz) were played back using a small MP3 player and
speaker. Not to scale. a and b were assigned to left or right side at
random for each maze

Oecologia
1 3
pre-selected side. The acoustic environment in the maze in
each treatment was measured using a Digitech QM-1589
sound meter lowered into the soil at the center of the maze.
As intended, sound levels in test treatments were 2–3 dB
greater than in the ZERO Hz control treatment (105 dB re
1 µPa).
Experiment 3: root responses to co‑occurring cues
The aim here was to evaluate the extent to which roots grow
in the direction that leads to water using acoustic informa-
tion when other cues occur simultaneously (as is likely in a
natural environment). We further examined the directional
behavior of roots in response to three test scenarios (TS7–
TS9; Table 1). In all three scenarios, the acoustic treat-
ment applied to one side of the maze (A) consisted of the
recorded sound of water [WTR-REC]. Depending on the
scenario, another treatment was concomitantly applied to
the other side (B). Specifically, in TS7 we used the pres-
ence of actual water contained in the transparent plastic
tray at the base of the maze [WTR] to test how roots use
sensory information delivered by moisture and acoustic
gradients to seek water when both are available. In TS8,
we applied the computer-generated white noise [NOISE]
to test whether roots could discriminate between different
sounds. And finally in TS9, we used the sound equipment
broadcasting no sound [ZERO Hz] to test whether other
factors that are due to the presence of the sound equipment
(e.g., effects of magnetic fields), affect or interfere with the
directional responses of roots to sound. As above, A and B
were assigned to left or right side at random for each maze.
Data analysis
As an initial test of overall significance, we used a propor-
tion test to evaluate whether there were any overall differ-
ences among the nine test scenarios in terms of their A vs
B proportions. We used a second proportion test to evaluate
whether the results were significantly different to random
[p(A) = p(B) = 0.5]. To fully analyze and differentiate the
various effects acting and interacting in each test scenarios,
we also conducted a detailed integrated analysis across
all nine test scenarios considered in the present study (see
details in electronic supplementary material). We fitted
binomial generalized linear models with no intercept terms
to predict the probability that root growth direction was to
the A treatment side, with model terms to represent each
potential effect hypothesized to be influencing the direction
of growth. The potential effects considered were an effect
of water contact [water contact]; an effect of real water
being present with or without contact [water present]; a
specific effect of the sound of water [water sound]; a spe-
cific effect of recorded white noise [white noise]; an effect
of the sound equipment being present [equip], and an addi-
tional effect of the sound equipment playing [equip play-
ing]. In each scenario, each effect was hypothesized to be
acting towards or against growth to the A side, or to have
no influence because the effect was not present at all in that
scenario or because it was present equally on both sides
(see Table 1 for full details of which effect was assumed
to be acting on which side in each test scenario). A model
with all these effects was fitted, and then simplified using
standard step-wise selection based on Akaike Information
Criteria (AIC) values to ensure that only useful predic-
tors were retained in the final model. We also considered
a general effect of a continuous sound [sound] (Table 1),
but as the sound effect is the sum of the white noise and
water sounds effects, it was not possible to test these three
effects independently. Therefore, we just tested the effect
of replacing [water sound] and [white noise] by [sound] in
Fig. 3 Photographic record of the position of the primary root
recorded on day 5. Firstly a the two small black pots at the base of
the maze were removed to expose the position of the primary root
in the maze (indicated by the red arrow) and then b seedlings were
extracted from the maze to expose the primary root (indicated by the
red arrow) relative to the whole root growth (color figure online)

Oecologia
1 3
the model. All analyses were conducted in the R software
environment. For the sake of reproducibility, the R script
used for analysis is provided in the online supplementary
material.
Results
Experiment 1: root directional growth towards the
sound of water
In the baseline TS1, an actual water gradient was produced
by the presence of 100 ml of water contained in the trans-
parent plastic tray at the base of one side of the maze (see
Table 1; Fig. 2a). In this scenario, 8 out of 10 (8/10) seed-
lings directed their root to the side of the maze where the
water source was located (Fig. 4). Seedlings in the TS2
were equally successful (8/10) at locating the water source
even though these seedlings had no direct access to water
but only to the live sound of water running inside a sealed
pipe (Fig. 4). The presence of circulating water in the pipe
had no effect on soil temperature (see Electronic Supple-
mentary Material for detailed results on temperature).
Experiment 2: root responses to recordings of natural
and artificial sounds
In TS3 where seedlings were played back the recorded
sound of running water, 6/10 of seedlings directed their
roots away from, rather than towards, the side of the maze
where the sound of water was located (Fig. 4). In TS4
where seedlings were exposed to computer-generated white
noise, the observed frequency of seedlings directing their
root away from the sound source increased to 8/10 (Fig. 4).
The avoidance behavior observed in TS3 and TS4 was
further intensified in TS5 (Fig. 4). In this scenario, 9/10
seedlings directed their roots away from the location of
the sound equipment that, albeit being turned on and play-
ing, broadcasted no actual sound. In TS6, when the sound
equipment was turned on but not playing [NOT PLAY-
ING], 6/10 seedlings grew away from the equipment
(Fig. 4).
Experiment 3: root responses to co‑occurring sounds
In TS7, where both moisture and acoustic gradients were
available, 2/10 seedlings directed their roots towards the
WTR-REC treatment side of the maze, 50% less than in
TS3 where the recorded sound of water was the only treat-
ment applied (Fig. 4). The number of seedlings that grew
to the WTR side where the water was physically present
in TS7 (8/10) was the same as observed in the TS1 sce-
nario where the WTR was the only treatment applied. In
the TS8 scenario where WTR-REC and NOISE treatments
were co-occurring, the number of seedlings that directed
their roots towards the side of the maze with the NOISE
(6/10) was threefold more than in the TS4 scenario where
the NOISE treatment was applied alone (Fig. 4). However,
the number of seedlings growing toward the WTR-REC
in this TS8 scenario (4/10) was the same as that observed
in the TS3 scenario where the WTR-REC was the only
Fig. 4 Number of seedlings
that directed their roots towards
the treatment side A of the
maze (white bars) across all test
scenarios (TS1–TS9; defined in
Table 1). The grey bars indicate
seedlings that did not choose
the treatment side. The red dot-
ted lines are intended to visually
distinguish the scenario specifi-
cally tested in each experiment.
See also Table S1 (color figure
online)

Oecologia
1 3
treatment applied (Fig. 4). And lastly, in the TS9 scenario
where WTR-REC and ZERO Hz treatments were applied at
the same time, the proportion of seedlings growing towards
the WTR-REC (7/9) was almost twice that observed in TS3
(4/10) when the treatment was applied alone (Fig. 4). The
proportion of seedlings growing toward the ZERO Hz in
this scenario (2/9) was also greater than that observed in
the TS5 scenario (1/9) where the ZERO Hz was the only
treatment applied (Fig. 4).
Integrated analysis of root responses
The proportion tests gave strong support that there were
overall differences among scenarios (p = 0.003), and
that results were not random (p = 0.002). As expected,
treatments had no effect on root L vs R growth direction
(p = 0.95) and the L vs R direction did not vary between
scenarios (p = 0.91). According to the integrated analy-
sis, the presence of the broadcasting sound equipment
had a strongly negative (repulsive) effect on root growth
(ΔAIC = 13.5), while contact with water (ΔAIC = 1.7),
the sound of water (ΔAIC = 6.9) and white noise
(ΔAIC = 1.7) all had positive (attractive) effects on root
growth, with the sound of water having the strongest effect
(Table 2). The potential hypothesized effects of presence of
real water and presence of the sound equipment were not
retained in the final simplified model, indicating no evi-
dence that these effects existed (see Table 2 for details of
the simplified model). Replacing the terms for water sound
and white noise with a single general sound effect did not
significantly reduce the explanatory value of the model,
indicating that the difference between the two types of
sound was not significant (p = 0.91).
Discussion
Our results demonstrate that garden pea seedlings respond
to acoustic vibrations generated by water moving inside
pipes and propagated through the substrate. Specifically,
results from Experiment 1 demonstrate that peas display
their typical hydrotropic behavior by growing their roots
towards the perceived water source even in the absence
of substrate moisture. Thus, it is not necessary for plant
roots to have direct access to moisture gradients to sense
that water is in the vicinity. Our results not only demon-
strate that roots can equally detect and use moisture or
acoustic cues to locate water, but also show that roots can
be selective about which cue is most advantageous in what
circumstance. When both gradients are accessible as in
Experiment 3, for example, roots preferentially use mois-
ture in the soil over acoustic vibrations. Given that acous-
tic vibrations propagate rapidly through soil, conveying
real-time information that can be analyzed quickly, sensed
at very low intensity and long distances, sound can be an
effective cue when a rapid and accurate decision about the
most effective direction of growth is required. Accordingly
and based on the present findings, we propose that acous-
tic gradients enable roots to broadly detect a water source
at a distance and conceivably, establish the most direct and
cost-effective route to that source prior to encountering the
associated moisture gradient. Once accessible, the mois-
ture gradient helps the roots to hone in on their target more
accurately, hence locating the exact location of the water
source.
Unexpectedly, seedlings directed their roots away from
the sound of water, when this was a recording played back
as in our Experiment 2. As a matter of fact, we found
that roots generally grew away from the side of the maze
where we positioned the sound equipment, regardless of
the broadcasted sound. Moreover, we observed this avoid-
ance response even when silence was played. One possible
explanation for these responses is that seedlings were able
to detect some other cue emitted by the sound equipment
(e.g., magnets in speakers), which affected their root direc-
tional growth. Because plants sense and integrate multiple
physical parameters for a range of tropic responses includ-
ing magnetotropism (Galland and Pazur 2005), we consid-
ered the possibility that the sound equipment we used in the
playback experiments emitted a magnetic field that seed-
lings sensed and selectively avoided. Hence, we tested the
sound equipment and measured a mean magnetic flux den-
sity of 3.7 nT ± 0.04 (SE), an intensity that was twice as
strong when compared to background readings (i.e., equip-
ment completely switched off; see Fig. S2 and details in
Electronic Supplementary Material). While it was beyond
the scope of this study to specifically test for magnetosen-
sory abilities of seedlings (see review by Maffei 2015), our
findings suggest that the sound equipment we used in the
Table 2 The final simplified model explaining the observed results
in terms of various effects potentially interacting in the different test
scenarios, showing for each effect (term retained in the simplified
final model), the estimated value for the model coefficient, the asso-
ciated p value, and the associated estimated probability that the root
would grow towards the side of the maze where only that effect is
acting, versus a side where no effect is acting
Final simplified model
Effect/model term Coefficient estimate p value Probability
estimate
Equip on −2.07 0.00 0.11
Water sound 1.36 0.01 0.80
White noise 1.30 0.06 0.79
Sound – – –
Water contact 1.07 0.07 0.74

Oecologia
1 3
playback experiments likely affected seedlings’ responses
to the test acoustic cues by interfering magnetically. Expo-
sure of pea seedlings to very low magnetic fields is known
to affect the ultrastructure of root cells by disrupting met-
abolic systems such as Ca2+ homeostasis (Belyavskaya
2001). Therefore, it is possible that the seedlings sensed
the magnetic output of our sound equipment as an envi-
ronmental condition to be avoided by moving away from
its source, even when another sought after cue (e.g., sound
of water) is delivered by the same source. This finding is
interesting because it demonstrates that seedlings have the
ability of “parsing” their sensory world into its components
of different types and hence, resolve the influx of informa-
tion by prioritizing cues that support the overall most ben-
eficial growth decision.
When the strong repulsive effect on root growth associ-
ated with the presence of the operating sound equipment
was experimentally standardized as in part of Experiment
3 (i.e., where individual peas were exposed to the magnetic
disturbance from both sides of the maze) and accounted for
in the integrated statistical model, there was some indica-
tion that the attractive effect of the recorded sound of water
was stronger than that of white noise. This would agree
with previous studies, which have demonstrated that plants
respond to vibrations in a selective way (Gagliano et al.
2012a; Appel and Cocroft 2014). By showing that Arabi-
dopsis plants were able to discriminate between the vibra-
tions caused by insect feeding and those caused by wind or
insect song, for example, Appel and Cocroft (2014) dem-
onstrated that the ability of plants to detect and selectively
respond to vibrations has an ecological function. In nature,
this selectivity in regards to sounds or vibrations could
explain how trees are able to detect different water sources
and discriminate between them based on their longer-term
availability. For example, streamside trees like scrub oaks
and box elders are known to preferentially tap into deeper
soil layers for a more reliable source of water rather than
the shallow streams, which are more ephemeral (Dawson
and Ehleringer 1991).
In our study, the difference between the sound of water
and white noise was not great enough to be significant,
which may be because the ability of seedlings to grow in
the direction that leads to water by using the sound of water
as a cue was contingent on the surrounding soundscape.
In both cited studies above (Gagliano et al. 2012a; Appel
and Cocroft 2014), plants were exposed to a single cue at
any given time, thus eliminating the possibility for acoustic
interference. In this study, however, seedlings in Experi-
ment 3 were subjected to two acoustic cues at once. It is
possible that the ability of seedlings to detect the recorded
sound of water was reduced because the cue was obscured
by the white noise (i.e., masking effect). Experimental stud-
ies in animal systems have demonstrated that the presence
of loud white noise can interfere with an individual’s abili-
ties to receive, respond and dispatch acoustic cues and sig-
nals. For example, Montgomerie and Weatherhead (1997)
showed that foraging success of American robins was
reduced when auditory cues were masked by white noise.
Bats and squirrels are also affected by white noise and
allocate little foraging time to these environments (Schaub
et al. 2008). If plant’s abilities to perceive and respond to
the surrounding soundscape are also affected by noise, as
our findings suggest, what are the ecological ramifications
of acoustic pollution on their natural communities? While
the urgent need to understand the consequences of altered
acoustics on animal populations has been increasingly rec-
ognized (Francis and Barber 2013), our findings clearly
indicate that the scope of our understanding on the matter
needs to be extended to include plants.
A better understanding of the acoustic ecology of plants
can also offer insights into new innovative practical appli-
cations. For example, our study clearly demonstrated that
plants are able to use sound to locate water inside sealed
pipes. We propose that some kind of adaptation to other
ubiquitous disturbances like those described in Gagli-
ano et al. 2012b (i.e., the resonant acoustic-free oscilla-
tions known as the Earth’s “hum”) and Appel and Cocroft
2014 (i.e., feeding sounds of herbivores) which also act
via vibrations may have been the key to the root behavior
described here. From an evolutionary perspective, the abil-
ity to respond to vibrations of various kinds including, for
example, those produced by a running stream would have
been highly relevant to a broad range of species and ben-
eficial to their survival. We suggest that plants already had
the ability to use information of vibrational origin by the
time humans started building their first underground pipes,
which archeological evidence dates to the Minoan civili-
zation of Crete during the Bronze Age from 3650 to 1400
BC (Wald 2016). Accordingly, plants have had millennia
to evolve this ability of responding to environmental vibra-
tions, including the sound of water moving through pipes.
This could explain why roots are particularly good at find-
ing and invading sewer pipe systems, even when their pipe-
lines are otherwise sealed and intact. Far from being trivial,
root invasion of sewer pipes has severe economic, environ-
mental and social consequences and is a major problem
for urban areas around the world. From 2006 to 2013, the
Water Corporation in Western Australia spent over AU$ 18
million on sewer pipe blockage repair, of which more than
65% were due to roots intrusion (Xie et al. 2014). In Ger-
many, the costs of root removal and associated pipe repairs
are estimated at EUR 28.4 million per year (Östberg et al.
2012). And boasting one of the largest and most complex
wastewater systems in the world approaching 7000 km of
sewer lines, the city of Los Angeles in the year 1997 spent
about US$ 5000 per km to remove roots from its pipes

Oecologia
1 3
(United States Environmental Protection Agency 1999). As
pointed out by Östberg et al. (2012), chemical applications
are considered to be the most efficient treatment against
root intrusion in pipes, but such an approach can only be
applied once the invasion has occurred and importantly, is
environmentally hazardous and not sustainable given that
its effects are not permanent. Yet, an alternative and envi-
ronmentally sustainable way to overcome problems asso-
ciated with root invasions may be as simple as utilizing
soundproof materials in the construction of sewer pipelines.
While the mechanisms of how plants detect sounds are
still to be identified (although possible mechanisms have
been suggested; Gagliano et al. 2012b, c; Gagliano 2013a,
b), plant responses to acoustic cues, whether these are
vibrations generated by neighboring plants (as suggested
by Gagliano et al. 2012b, c), attacking herbivores (Appel
and Cocroft 2014), buzz pollinators (Proctor et al. 1996) or
human activities (as shown in this study), demonstrate that
sound and vibrations play an important ecological role in
the life of these organisms. Hypothesis-driven research is
required to systematically investigate the capacity of plants
to detect and use sounds. The key questions to be addressed
are what are the mechanisms underlying the ability of
plants to discriminate sound sources and their informa-
tion content, and how this knowledge can be responsibly
applied in a beneficial manner to plants and animals alike.
The answers are clearly important to better understand the
processes underlying species interactions and co-evolution,
including bio-inspired innovative solutions for application
to real world problems.
Acknowledgements We thank R. Creasy, W. Piasini, H. Etchells, T.
Betts, N. Clairs, R. Malkin and P. Tallai for their assistance, and H.
Heilmeier and two anonymous reviewers for valuable comments on
the manuscript. This work was supported by Research Fellowships
from the University of Western Australia and the Australian Research
Council (ARC grant n. DE130100018) to MG*.
Author contribution statement MG* conceived and designed the
experiments. MG* and MG performed the experiments and collected
data MG*, MD and MR analyzed and interpreted the data. MG* and
MR drafted the paper. All authors edited and critically revised the
final version, and approved its publication.
Compliance with ethical standards
Conflict of interest The authors declare no competing interests.
References
Appel HM, Cocroft RB (2014) Plants respond to leaf vibrations
caused by insect herbivore chewing. Oecologia 175:1257–1266.
doi:10.1007/s00442-014-2995-6
Bekele T, Olsson K, Olsson U, Dahlborn K (2013) Physiological and
behavioral responses to different watering intervals in lactating
camels (Camelus dromedarius). Am J Physiol Regul Integr
Comp Physiol 305:R639–R646
Belyavskaya NA (2001) Biological effects due to weak magnetic
field on plants. Adv Space Res 34:1566–1574. doi:10.1016/j.
asr.2004.01.021
Bernáth B, Gál J, Horváth G (2004) Why is it worth flying at dusk
for aquatic insects? Polarotactic water detection is easiest at
low solar elevations. J Exp Biol 207:755–765. doi:10.1242/
jeb.00810
Bielenberg DG, Miller JD, Berg VS (2003) Paraheliotropism in
two Phaseolus species: combined effects of photon flux den-
sity and pulvinus temperature, and consequences for leaf gas
exchange. Environ Exp Bot 49:95–105
Cassab GI, Eapen D, Campos ME (2013) Root hydrotropism: an
update. Am J Bot 100:14–24. doi:10.3732/ajb.1200306
Dawson TE, Ehleringer JR (1991) Streamside trees that do not use
stream water. Nature 350:335–337. doi:10.1038/350335a0
United States Environmental Protection Agency Office (1999) Col-
lection systems O & M fact sheet: sewer cleaning and inspec-
tion. Publication EPA 832-F-99-031, Washington DC. http://
water.epa.gov/scitech/wastetech/upload/2002_06_28_mtb_
sewcl.pdf
Francis CD, Barber JR (2013) A framework for understanding noise
impacts on wildlife: an urgent conservation priority. Front Ecol
Environ 11:305–313
Gagliano M (2013a) Green symphonies: a call for studies on acoustic
communication in plants. Behav Ecol 24:789–796
Gagliano M (2013b) The flowering of plant bioacoustics: how and
why. Behav Ecol 24:800–801
Gagliano M, Mancuso S, Robert D (2012a) Towards understanding
plant bioacoustics. Trends Plant Sci 17:323–325
Gagliano M, Renton M, Duvdevani N, Timmins M, Mancuso S
(2012b) Acoustic and magnetic communication in plants: is it
possible? Plant Signal Behav 7:1346–1348
Gagliano M, Renton M, Duvdevani N, Timmins M, Mancuso S
(2012c) Out of sight but not out of mind: alternative means of
communication in plants. PLoS One 7:e37382. doi:10.1371/jour-
nal.pone.0037382
Galland P, Pazur A (2005) Magnetoreception in plants. J Plant Res
118:371–389. doi:10.1007/s10265-005-0246-y
Hart JW (1990) Plant tropism and growth movement. Unwin Hyman,
London
Hawkins BA et al (2003) Energy, water, and broad-scale geographic
patterns of species richness. Ecology 84:3105–3117
Jaffe MJ, Takahashi H, Biro RL (1985) A pea mutant for the study of
hydrotropism in roots. Science 230:445–447
Kiss JZ (2007) Where’s the water? Hydrotropism in plants. Proc Natl
Acad Sci USA 104:4247–4248. doi:10.1073/pnas.0700846104
Kiss JZ, Millar KDL, Edelmann RE (2012) Phototropism of Arabi-
dopsis thaliana in microgravity and fractional gravity on the
international space station. Planta 236:635–645. doi:10.1007/
s00425-012-1633-y
Ledger ME, Brown LE, Edwards KE, Milner AM, Woodward G
(2013) Drought alters the structure and functioning of complex
food webs. Nature Clim Change 3:223–227
Maffei ME (2015) Magnetic field effects on plant growth, devel-
opment, and evolution. Front Plant Sci 5:445. doi:10.3389/
fpls.2014.00445
McCluney KE, Sabo JL (2009) Water availability directly deter-
mines per capita consumption at two trophic levels. Ecology
90:1463–1469
McCluney KE et al (2012) Shifting species interactions in terrestrial
dryland ecosystems under altered water availability and climate
change. Biol Rev 87:563–582
Montgomerie R, Weatherhead PJ (1997) How robins find worms.
Anim Behav 54:143–151

Oecologia
1 3
Niklas KJ (1997) The evolutionary biology of plants. The University
of Chicago Press, Chicago
Östberg J, Martinsson M, Stål Ö, Fransson A-M (2012) Risk of root
intrusion by tree and shrub species into sewer pipes in Swedish
urban areas. Urban For Urban Green 11:65–71. doi:10.1016/j.
ufug.2011.11.001
Proctor MCF, Yeo P, Lack A (1996) The natural history of pollination.
Timber Press, Portland
Russell J, Vidal-Gadea AG, Makay A, Lanam C, Pierce-Shimomura
JT (2014) Humidity sensation requires both mechanosensory and
thermosensory pathways in Caenorhabditis elegans. Proc Natl
Acad Sci USA 111:8269–8274. doi:10.1073/pnas.1322512111
Schaub A, Ostwald J, Siemers BM (2008) Foraging bats avoid noise.
J Exp Biol 211:3174–3180. doi:10.1242/jeb.022863
Scott P (2000) Resurrection plants and the secrets of eternal leaf. Ann
Bot 85:159–166
Wald C (2016) The secret history of ancient toilets. Nature
533:456–458
Wolverton C, Kiss JZ (2009) An update on plant space biology. Gravit
Space Res 22:13–20
Xie Q, Hodkiewicz MR, Khan N, Best A (2014) Predictive model-
ling of sewer blockages in vitrified clay pipes. CEED Seminar
Proceedings