Photosynthesis by marine algae produces sound, contributing to the daytime soundscape on coral reefs


We have observed that marine macroalgae produce sound during photosynthesis. The resultant soundscapes correlate with benthic macroalgal cover across shallow Hawaiian coral reefs during the day, despite the presence of other biological noise. Likely ubiquitous but previously overlooked, this source of ambient biological noise in the coastal ocean is driven by local supersaturation of oxygen near the surface of macroalgal filaments, and the resultant formation and release of oxygen-containing bubbles into the water column. During release, relaxation of the bubble to a spherical shape creates a monopole sound source that ‘rings’ at the Minnaert frequency. Many such bubbles create a large, distributed sound source over the sea floor. Reef soundscapes contain vast quantities of biological information, making passive acoustic ecosystem evaluation a tantalizing prospect if the sources are known. Our observations introduce the possibility of a general, volumetrically integrative, noninvasive, rapid and remote technique for evaluating algal abundance and rates of primary productivity in littoral aquatic communities. Increased algal cover is one of the strongest indicators for coral reef ecosystem stress. Visually determining variations in algal abundance is a time-consuming and expensive process. This technique could therefore provide a valuable tool for ecosystem management but also for industrial monitoring of primary production, such as in algae-based biofuel synthesis.


Coral reef degradation and algal smothering are caused by human impacts such as overfishing, pollution through nutrient runoff and climate change effects. Algal dominance diminishes the value of reefs [1], is likely permanent [2, 3], and serves as a clear indicator of ecosystem degradation [4]. Timely monitoring of reef state is crucial to quantifying impacts and mitigating ecosystem shifts [5], but present-day assessment methods are resource intensive and infrequent [6]. Because coral reefs are highly soniferous ecosystems where most sound comes from biological processes, ecologists have sought to apply the integrative acoustic monitoring approaches used successfully in terrestrial soundscape ecology [7, 8]. These techniques can yield spatially integrative and long-term continuous observations that can be collected autonomously with inexpensive recorders, but the sources of many reef sounds are challenging to identify [9]. Previously we established positive correlations (Pearsons ρ) between benthic macroalgal cover and aspects of coral reef soundscapes (acoustic pressure spectral density levels in the 2 to 20 kHz band) at 17 sites over a 2570 km transect from Kure Atoll to the Island of Hawaii [10]. These correlations were only observed during the day when overall sound levels were typically lower (Table 1), and correlations of this nature did not exist between sound and any other non-acoustic environmental metric. To further deconvolute the soundscape sources, we performed tank-based passive acoustic experiments with the invasive Hawaiian algae Salicornia gracilaria. Ambient sound was recorded in combination with dissolved oxygen and time-lapse photography of algae under daylight and night conditions to investigate the mechanism behind the observed associations of sound with algal presence.


Table 1. Correlation between benthic algal cover and coral reef soundscapes.

Pearson’s correlation coefficients between the percentage of benthic macroalgal cover at 17 reef sites throughout the Hawaiian Islands and ambient acoustic pressure spectral density in three bands between 2 and 20 kHz, obtained simultaneously at the same locations. Intensity-filtered spectra were averaged over one hour beginning at the indicated times.

Short-time Fourier analysis on intensity filtered soundscape recordings [10] provide a more spectrally detailed view of individual transient soundscape components (i.e., most reef-based biological sound). Typical Fourier analysis approaches in underwater acoustics integrate over time periods that are much longer than each individual biological sound. While longer integration results in increased frequency resolution, ideal for detecting tonal and/or narrow-band sources, the approach can also spectrally smear multiple transient sounds together and/or reduce peak level estimates through giving equal weighting to quiet periods between transient arrivals. If a persistent environmental noise is present, it can mask some transient biological sounds sampled through this approach. Our approach involved selecting each transient using an intensity filter, ensuring the transform length encompassed only the transient, and then assessing the spectral qualities of each transient individually. As a consequence, reef spectra shown here cannot be directly compared to spectra obtained without intensity filtering. Depending on ambient and preamplifier noise levels, lower-level sounds from individual biological sources may not be discernable at significant distances from the source. However, many of these events occurring over a distributed area create an extended sound source for which propagation models show decreased attenuation with distance [11].


With the onset of photosynthesis during light periods, bubbles could readily be observed with the naked eye on the surface of macroalgae (Fig 1A and 1B). As the bubbles detached from the plant, they created a short ‘ping’ sound. Acoustic recordings of bubble release separated by quiet periods consequently appear as an irregular pulse-train-like time series (Fig 1C). The sound created by perturbation of the bubble during release is naturally transient and decays exponentially (Fig 1D). Oversampled recordings (100 kS·s-1) permit more detailed spectral analysis of each sound (Fig 1E) and fundamental frequency estimates for each bubble can be made from the resultant data (Fig 1F). Tank resonance modes [12] were restricted through the presence of algae and the drafted, acoustically absorptive nature of the tank walls and did not appear to substantially influence the recorded spectra (Fig 1G).


Fig 1. Characterization of bubble sound from algae.

(A) Gracilaria salicornia actively creating gas bubbles during photosynthesis. White arrows indicate the locations of larger bubbles. The scale bar is 5 mm in length (B) A closer view of one gas bubble about to detach from the algae surface. The scale bar is 2 mm in length. (C) A 9 s time series of ambient sound from inside the aquarium. (D) A high-resolution view of a typical transient waveform as shown in (C). (E) Spectrogram showing the time-varying spectral content of the received waveform in (D). (F) Pressure spectral density (PSD) estimates showing mean, median and maximum pressure spectral densities of received level from the 2 ms period as shown in (D) and (E), indicating the spectral peak of the waveform as 13.07 kHz (vertical dashed line). (G) An averaged spectrum from transient sounds recorded over one hour overlaid with standing-wave resonant tank modes. Mode sums indicate the sum of mode numbers in the horizontal (length, width) and vertical (depth) directions.

Experiments with S. gracilaria in controlled laboratory settings demonstrate that sounds produced by algal photosynthesis are similar in nature to components of the soundscape recorded in shallow water regions where S. gracilaria and other benthic macroalgae species are common. With the intermittent application of Photosynthetically Active Radiation (PAR, μ = 381, σ = 67 μmol·m-2s-1, roughly equivalent to the available level at water depths of 10 m on a cloudy day in the equatorial pacific) the link between photosynthesis, dissolved oxygen and acoustic emissions through bubble formation becomes apparent (Fig 2 and S1 Fig). Over the duration of PAR application both the number of bubbles (R2 = 0.76, S2A Fig) and their mean size (R2 = 0.47, S2B Fig) increased with dissolved oxygen levels. The process was reversed with the removal of light. These increases in the size and rate of bubble formation during the illuminated period lower the frequency distribution of sounds produced by the algae (Fig 2A) and cause an increase in the Sound Exposure Level (SEL, R2 = 0.82, Fig 2B and S2C Fig), a measure analogous to acoustic work. The onset of bubbling rate and size increase from the application of light was delayed due to the time required for photosynthesizing algae to build supersaturation conditions in the surrounding water, and for bubbles to reach a size large enough to create sufficient buoyancy force to detach from the algae. After removal of the light source there was a comparable delay in the reduction of bubble production and sound, due to bubbles remaining on the algae and saturation state remaining above 100 percent.


Fig 2. The acoustic response of algae to light over time.

(A) Ten-minute averaged time-frequency histograms showing the distribution of Minnaert frequencies from bubbles produced by algae with the application and removal of PAR. The color scale indicates the number of bubble events acoustically characterized over a 10-min period, per frequency bin (195 Hz bin width). (B) Sound Exposure Level (SEL, 10 min. averages, ± 1 S.E., left axis) and dissolved oxygen concentration (right axis) showing an increase, decrease, and increase of dissolved oxygen and SEL with the application, removal and application of PAR, respectively. The grey regions indicate the period when the light source was removed.

A comparison between acoustically derived bubble size distributions and photographically obtained measurements of bubble size (Fig 3) optically validates the acoustic estimation of algae-driven bubble radii (99 percent significance: no evidence the distributions were unequal, Wilcoxon 2-sided test, p = 0.9087).


Fig 3. Size distribution comparisons between bubble radii distributions simultaneously derived through photography and passive acoustic recording.

(A) Bubble size distributions obtained from photographically imaged measurements (histogram, left axis) and Minnaert radii (solid line, right axis) every 150 s over a one-hour period of active bubble formation. (B) Boxplots indicating 25th, 50th, 75th percentiles and 5th and 95th percentile outliers (crosses) of optically imaged and acoustically derived bubble radii distributions over the same period.


Algae release oxygen as a byproduct of photosynthesis. While these waste molecules are formed intracellularly in solution, the nucleation of oxygen gas bubbles on the surface of macroalgal tissue takes place when localized supersaturation of dissolved oxygen occurs at a nucleation site. Depending on the timescale of bubble formation and the total gas tension, the diffusion of nitrogen will also contribute to total bubble volume. Previous work has shown the relationship between oxygen bubble production by algae and oxygen supersaturation at their surface microenvironment [13]. The bubbles grow with the addition of more waste oxygen produced by photosynthesis and nitrogen diffusion, ultimately separating from the algae through a combination of buoyancy and surface tension forces. As a spherical bubble is perturbed through release, it oscillates in volume and pressure with exponential decay at a frequency inversely proportional to its radius R0, the roots of the specific heat ratio of the gas γ and ambient fluid pressure Pfl, and the inverse root of ambient fluid density, ρfl, a relationship first derived by Minnaert [14] and now referred to as the Minnaert frequency [15] ωMinn: (1)

This approximation was first developed with the assumption of negligible heat flow (adiabatic conditions) and negligible surface tension, but has since been shown to be a good approximation for bubbles of radii between 30 nm to 300 μm. Consequently, passive acoustic estimates of bubble volume can be made if the water properties and depth at which bubble separation occurs are known.

Our observations demonstrate the mechanism behind previously established correlations between components of underwater soundscapes and the relative abundance of macroalgae on Hawaiian coral reefs [10]. Although ecologically distinct, bubble production is ubiquitous in sea grass beds [16], and can also be observed in marine algal species [17]. Thus, the bubble production mechanism is not specific to S. gracilaria and may be used as a general indicator of photosynthetic activity. Ultrasonic emissions from terrestrial plants are also driven by bubble processes, but the underlying mechanism is water starvation rather than photosynthesis [18].

The contribution of algal sound to soundscapes may be isolated from other ubiquitous sound sources though matched filtering, directional receivers and similar techniques. The observed size distribution and thus corresponding spectral distribution of bubble sound (Fig 3A), and the a-periodic and un-clustered nature of acoustic emission time series (Fig 1), are not shared by any other bubble related process in shallow-water environments, including breaking waves [19]. While active acoustic techniques have been used to estimate marine plant biomass [16, 20], our finding introduces the possibility of noninvasive sensing methods that can quantify relative primary production rates in addition to presence/absence, although further work in disentangling confounding factors is necessary. The relationship between bubble formation processes and gas transfer mechanisms has been well studied [13, 2122] providing a background for the application of bubble science to underwater ecological sensing.

The findings presented here show that algae are capable of producing sound under normal circumstances and that sound may explain a correlative association previously discovered between sound in the 2 to 20 kHz band and algal cover on coral reefs [10]. However, a great deal of further work is required before a passive acoustic tool could be developed for quantifying bubble output from photosynthesis, in-situ primary productivity and algal abundance. A number of caveats in our analysis must be considered and results from soundscape correlation and tank experiments should be considered limited in scope until these caveats are resolved. Underwater soundscapes, especially in the vicinity of a shallow coral reef, are extremely complicated acoustic environments containing sound sources of many types [23]. A number of spatio-temporally variable physical mechanisms that influence the production and propagation of sound from source to receiver also exist.

The most notable interference may be snapping shrimp noise. While the sound generation mechanism of these shrimp involves cavitation [24] and is thus spectrally unique from the relatively narrow-band bubble emissions discussed here, source levels and rate of occurrence can be so high [2527] that they are capable of ‘drowning out’ other sound sources for protracted periods of time. Furthermore, the spectral, temporal frequency, and diurnal behavior of snapping shrimp vary depending on the location and time of year [2829, 2627]. While the frequencies produced by photosynthetic bubbles are within the band of sound produced by snapping shrimp, spectral analysis of sound from shrimp-dominated and coral reef soundscapes reveal spectral differences that suggest other biological contributors can be spectrally evaluated even in the presence of shrimp noise. Bubbles produce transient sounds that contain energy in a relatively narrow frequency band. Due to the higher likelihood of smaller bubbles, it is less likely that bubbles would add significant energy at lower frequencies (i.e., below 5 kHz). Conversely, sound produced by snapping shrimp is broad-band but weighted toward a peak between 4 to 6 kHz [25]. The integration of snapping shrimp sounds over time can produce a low-frequency-weighted spectrum with monotonic decay. Conversely, some reef soundscapes may create higher-frequency-weighted spectra that may include distinct spectral peaks (Fig 4A). Limited comparisons between a snapping-shrimp-dominated soundscape and those correlated with algal dominance in Table 1 show a variation in band levels between 5 and 20 kHz between sites. In this case, the shrimp-dominated soundscape is the Scripps Pier in La Jolla, Ca. Periodic cleaning of encrusting sponges on the pilings reveals large communities of snapping shrimp. Care should be taken when evaluating this comparison as the behavior and sound of snapping shrimp may vary between Hawaii and San Diego. However, the Scripps Pier is relatively unique in that no reefs exist within half a mile of the pier. Recordings were made on a calm day with little swell. The comparison shows that in daytime reef soundscape levels were between 5 to 12 dB higher above approximately 5 kHz. Barring differences in frequency-dependent attenuation or shrimp sound characteristics, the comparison suggests an additional source of higher frequency sound present at the reef sites (Fig 4B). The high source levels emitted by snapping shrimp drive the prevailing thought that they are the overwhelmingly dominant bioacoustic source in coastal underwater ecosystems. However, our suggestion that individually quieter biological sources of sound are detectable in the presence of snapping shrimp is not unprecedented. Chorusing from sea urchin grazing is known to contribute to soundscape spectra off the coast of New Zealand [30] and hermit crabs have been spectrally matched to soundscapes from environments in which both they and snapping shrimp are plentiful [9]. Nevertheless, before accurate quantification of algal photosynthetic activity can be attempted in the field, the behavior and acoustic characteristics of snapping shrimp in the area must be understood to create validated acoustic signal processing algorithms capable of differentiating between shrimp noise and algal sound.


Fig 4. Spectral comparisons between Hawaiian reef soundscapes and the snapping-shrimp dominated soundscape at the Scripps Pier in La Jolla, CA.

(A) Normalized, intensity filtered pressure spectral density levels recorded at midday local time during sunny days on shallow Hawaiian reefs (colored lines) plotted with similarly filtered levels obtained during crepuscular chorusing of snapping shrimp communities on the Scripps pier (black line with grey shading underneath). (B) Logarithmic spectral level differences between crepuscular pier and midday Hawaiian reef soundscapes in Fig 4A. Note that 3 dB represents a doubling or halving of the spectral level difference in each frequency bin. Hawaii locations are identified as follows: FFS–French Frigate Shoals; Kure–Kure Atoll; Big I.–Ke’ei Beach, Big Island Hawaii; Lis.–Lisianski Island; Oahu–Lai’e Beach, Oahu, Hawaii; Maui–La Perouse Bay, Maui, Hawaii; PHR–Pearl and Hermes Reef; Kauai–Tunnels Beach, Kauai, Hawaii.

Another confounding factor may be the influence of water movement on bubble formation and retention. Bubbles will not form if water is continually swept from the algae and oxygen saturation state never exceeds 100 percent. Secondly, if bubbles are formed, they may be prematurely removed through wave action, decreasing the size distribution of the bubbles and shifting the frequency distribution higher. The mechanisms governing benthic oxygen saturation state and the effects of water movement on bubble retention are poorly understood. Benthic algae reside at least partially within the fluid boundary layer, especially in rugose environments, and may be somewhat protected from flow. The structure of algal filaments may also assist or retard the removal of bubbles by wave action, meaning that aspects of sound production are likely species-specific. While bubble retainment by algae in current remains to be quantitatively investigated, Fig 5 shows typical turf algae in a shallow reef environment off Hawaii’s Big Island on a sunny day. Fig 5 demonstrates that in very shallow water subject to wave action, bubbles from photosynthesis may be retained if sufficient algal structure exists. Further acoustic, flow, and bubble physics analysis, along with an investigation of biological interaction (both floral and faunal) are required to quantify and understand the impact water movement has on bubble formation and the consequent acoustic emissions by algae in these scenarios.


Fig 5. Bubbles from photosynthesis present on shallow algal turf subject to swell.

Note the movement of suspended particles during the camera exposure period of 1/125 sec. Scale bar indicates 2 cm. Image courtesy Florybeth La Valle, University of Hawaii.

The probability that too many transient sounds arrive at the receiver during one FFT integration period for spectral analysis on individual sounds to be performed increases with the size of the reef, the nature of propagation pathways from sources to the receiver, and the acoustic productivity of the benthic community. As the range between any given source and the receiver increases, the probability that multipath arrivals are interpreted as separate signals also rises. However, the spectral content of multipath arrivals is less likely to differ significantly from the direct arrival when compared to other sound sources. Future data collection with a spatial filter such as a directional receiver or hydrophone array can limit the area over which passive acoustic surveys are performed, simplifying the challenges associated with multipath and the large number of sources.

Sound is the most efficient radiation underwater and we continue to discover that important biological components of many aquatic ecosystems are both sensitive to [31] and produce acoustic emissions. Contributions to biological soundscapes on coral reefs include sound deliberately produced by fishes and invertebrates for communication [32] and defense [25], or inadvertently through feeding [30] and movement [9]. To date these sounds have been used to study the behavior and distribution of marine mammals and fish [3233], to understand ambient noise levels in the ocean [3435], and to assess the impacts of anthropogenic noise on marine environments [36]. Soundscape measurements are also effective at night when benthic activity is greatest [9] and passive optical techniques are difficult or impossible to implement [37]. In concert with these acoustic observations, a volumetrically integrative, remote and noninvasive method of measuring primary productivity outputs may lead to more accurate, persistent and less expensive surveys of ecological state in shallow water, the majority of which are currently performed in-situ through diver-based optical methods. Quantifying primary productivity through passive acoustic monitoring could become a technique that may not be limited to coral reef environments. Macroalgae are the dominant benthic primary producer in many temperate coastal and freshwater environments that are more difficult to survey optically due to reduced water clarity. Passive acoustic monitoring of ebullition has been utilized in the monitoring of chemical reactions [38]. A semi-real-time and volumetrically integrative approach can be adopted in biofuel-generating algal reactors using similar methods. The sensing technique proposed here is restricted neither to in-situ observations nor to marine environments but may apply basically to all aquatic environments in which macroalgae grow.

Materials and methods

Background data

The summary data shown in Table 1 were collected in the Northwest Hawaiian Islands (NWHI) during a NOAA Remote Areas Monitoring Program Cruise HA-12-04 on the R/V Hi’Ialakai during August 2012 and from sites accessible from shore on the Main Hawaiian Islands during September-October 2012. Permits were obtained for conducting research in the Papahānaumokuākea Marine National Monument (PMNM-2012-029) and the Main Hawaiian Islands (Department of Land and Natural Resources special activity permit 2012–83). Acoustic data were collected using a single hydrophone Loggerhead Instruments DSG Ocean recorder configured with a sampling frequency of 80 kSs-1 and an electronic gain of 20 dB. Simultaneous with acoustic recorder deployments, benthic phototransects were taken in accordance with NOAA Coral Reef Ecosystem Division rapid ecological assessment protocol [39]. Pearson’s correlation coefficient, ρ, between all phototransect-derived environmental variables, and band-limited pressure spectral density estimates, were calculated over data from all field sites for which acoustic and phototransect data were available. Metrics for which ρ values were greater than 0.6 with p < 0.001 (Bonferroni correction) were considered to be sufficiently correlated. Correlations of particular interest were those that appeared to form a distinct pattern across several metrics (i.e., similarly high correlation across adjacent acoustic bands, combined with a consistent temporal pattern). A full description of the data collection and processing methods associated with Table 1 is available in the associated publication [10].

Summary of the experiment

Tank-based experiments were carried out at the Hawaii Institute of Marine Biology at Coconut Island, Kaneohe Bay, Hawaii. Ten kilograms (wet) of the invasive red algae Gracilaria salicornia were collected by the Department of Aquatic Resources (DAR) of the State of Hawaii in Kaneohe Bay and stored in a 1 m diameter seawater holding tank with a steady filtered (100 um) ocean water flow. One kilogram of algae was visually inspected to remove any associated fauna and relocated to a smaller, plastic, opaque-sided rectangular tank with internal dimensions of 550 x 300 x 300 mm. Seawater was used to fill the tank to a depth of 250 mm and a 50 mm layer of algae was positioned on the floor of the tank between 200 and 250 mm depth. The algae remained negatively buoyant throughout the experiment. An aquarium light (Radion XR30w G4 pro) was positioned centrally over the tank and 228 mm (9 in.) above the water surface. All testing was conducted in a dark room with the aquarium light being the only source of photosynthetically available radiation (PAR). The color temperature was set at close to that of natural sunlight on a clear day (5500 K), at which the light operated at 55 percent of maximum output. Software limitations required a warm-up and cool-down period of 30 min before and after full-strength application, during which light levels were linearly ramped up and down between zero and 55 percent. Mean PAR level was obtained from measurements at fifteen equally spaced locations at algae equivalent depth using a calibrated LI-COR Li-193 spherical underwater quantum sensor and a LI-COR Li-250 light meter. Note that the mean PAR level obtained (381 μmol·m-2s-1) is far below what is expected near the surface on a calm sunny day in the tropics [40]. The experiment was started with algae being brought out of a 24-hour period of darkness and exposed to PAR as described above. After 3 h and 45 min, light levels were reduced to zero. After 7 h of darkness, light levels were increased back to the levels described above until the conclusion of the experiment after 10.5 h of light. The algae were then removed from the tank and disposed of in accordance with DAR procedure.

Acoustic recordings

Acoustic data were collected with a High Technology Inc. HTI-92-WB hydrophone, equipped with a low-power, high sensitivity preamplifier. Hydrophone and preamp system sensitivity was -144.8 dB re 1 VμPa-1 over the 2 Hz to 40 kHz band. The hydrophone was positioned to one side of the tank at 125 mm depth, 60 mm from the nearest tank wall. The preamplifier was powered by a variable DC power supply set at 10 V and drawing 2 mA. Data were acquired using a National Instruments® USB-6366 data acquisition module connected to a laptop computer running National Instruments® LabView® 2013 (SP1) and customized script. Time-stamped data were sampled at 100 KSs-1 at 16-bit depth in discrete 10 s intervals, pausing for a short time in between to save the data and verify that the sampling frequency was accurate. The recordings were processed in Mathworks Matlab® software. Before analysis, a high-pass, 5th order Butterworth filter (stop band: 0–0.5 kHz, pass band: 0.8–50 kHz was applied to remove DC offset, 60 Hz electrical noise or low frequency noise from passing vehicles and other sources.

Acoustic data processing

The peaks in amplitude associated with bubble separation ‘pings’ were identified from pre-filtered acoustic data if absolute peak amplitudes were greater than 5 σ from preamp self-noise levels. A 10 ms window around each of these peaks was isolated for further analysis. Information regarding the number of qualifying peaks and the temporal distribution of inter-peak periods within each 10 s file was retained. In order to calculate the sound exposure level (SEL) of bubble sound, the time duration of each air bubble “ping” was determined as the time between the 5th and 95th percentiles of the cumulative energy within the 10 ms window. SEL values were averaged over successive 10-min periods. Spectral analysis was performed on the time series between the 5th and 95th percentiles using a Kaiser-Bessel-windowed (β = 2.5π) 512-point FFT, overlapped 90 percent, meaning a bin width of 195.3 Hz and an overlapped spectral estimate made every 512 μs. The frequency bin containing the highest pressure spectral density level above 1 kHz, temporally matched to the amplitude peak in the corresponding time series, was determined to best represent the dominant frequency of each bubble separation ping. Regression of log-linear relationships and calculation of residuals was conducted using a simple logarithmic (base ten) model for the relationship between SEL and dissolved oxygen, and a natural logarithmic model for the relationship between the number and mean size of bubbles and dissolved oxygen. Residuals were calculated using linearized values.

Determination of bubble size

Estimates of bubble radii (R0) corresponding to each peak frequency were made using the Minnaert equation (1). For R0 to be a good approximation for the true radii of a freely oscillating bubble, the bubble radius must satisfy the following conditions: that the bubbles are “acoustically small”, the geometric mean of bubble radius and acoustic wavelength is large compared to the viscous boundary layer, the bubbles are “thermally large”, and the Laplace pressure is much less than the equilibrium pressure in the liquid (20).

Dissolved oxygen, pH conductivity, and temperature data

Dissolved oxygen (DO), pH, conductivity, and temperature were measured in 60 s intervals over the duration of the experiment. All non-acoustic sensors were part of a Manta+ 20 multi-probe system placed inside the experiment tank opposite the hydrophone. DO was assessed with an optical dissolved oxygen sensor provided by the Hamilton company (HDO) with a resolution of 0.01 mgL-1 or 0.1% saturation. pH was assessed via an electrolyte filled glass sensor next to a reference electrode. For conductivity, a four electrodes sensor was used while the temperature was assessed with a thermistor. Additional specifics are provided by the manufacturer [41]. While essential for the experiment, the oxygen sensor produced two types of acoustic noise that were additionally filtered from the acoustic data before analysis. The first type of noise was a startup transient that occurred upon system power-up every 60 s. A matched filter was used to exclude 0.2 s portions of recordings in which the transient was present. Secondly, a steady-state doubling of high frequency noise levels was emitted by the sensor for approximately 8 s after the startup transient. These recording periods were identified using a Hilbert-transform-based threshold detection algorithm and excluded from analysis.

Optical validation

Bubble radii were directly measured from high resolution photographs taken using a Canon SL1 single lens reflex camera equipped with a Canon 100 mm F2.8 L macro lens (160 mm effective) mounted on a tripod outside of the aquarium and aimed inward through a transparent window. A scale rule bar was inserted into the aquarium and placed across the field of view, orthogonal to the lens axis and within the depth of field. Only bubbles that were in focus within the depth of field of each image were analyzed. Photographs were taken every 150 s for one hour during a period when algae were observed to be actively producing bubbles. Radii measurements were obtained in post-processing using ENVI software, measured by pixel width and converted to μm using the scale bar in individual photographs.

Supporting information



1 / 2


Chemical and acoustic responses of algae to changes in photosynthetically available radiation (PAR).

All subfigures are time-aligned. (A) Dissolved oxygen time series. (B) Acoustically determined bubble counts per minute. (C) Acoustically determined mean bubble radius per minute.


S1 Fig. Chemical and acoustic responses of algae to changes in photosynthetically available radiation (PAR).

All subfigures are time-aligned. (A) Dissolved oxygen time series. (B) Acoustically determined bubble counts per minute. (C) Acoustically determined mean bubble radius per minute.


S2 Fig. Log-linear regression plots of acoustic emissions from photosynthesizing algae.

(A) No. of bubbles per minute against dissolved oxygen levels (R2 = 0.76, exponential coefficients α = -2.54, β = 0.43). (B) Mean bubble radii per minute against dissolved oxygen levels (R2 = 0.47, exponential coefficients α = -3.85, β = 0.19). (C) 10-minute Sound Exposure Level against dissolved oxygen levels (R2 = 0.82, exponential coefficients α = 83.15, β = 1.39). The coefficients may be applied to an exponential regression of linear parameters x and y as follows: (2)



We thank G. B. Deane, F. L. Rohwer, B. Bailey and R. F. Keeling for helpful comments and suggestions; K. Lubarsky, B. Neilson, D. Lager, K. Fuller, and K. Tucker for providing algae samples; J. Davidson and W. Au for providing laboratory facilities; S. Calhoun and B. T. Reyes for calibration of laboratory equipment; F. La Valle for the image in Fig 5; J. Tomlinson, P. D. Franck and S. M. Franck for experiment logistical support; G. J. Orris, S. Ackelson, R. D. Norris, M. J. Buckingham and A. J. Miller for funding support. This work was supported by NSF IGERT grant no. 0903551, U.S. Naval Research Laboratory section 219 program funds, National Research Council and American Society for Engineering Education postdoctoral fellowships and NASA grant no. NASA-14HYSP14-0003. Two anonymous reviewers provided comments that greatly improved the quality of the manuscript.


  1. 1. Haas AF, Guibert M, Foerschner A, Calhoun S, George E, Hatay M, et al. Can we measure beauty? Computational evaluation of coral reef aesthetics. PeerJ. 2015;3: pe1390. pmid:26587350
  2. 2. Scheffer M, Carpenter S, Foley JA, Folke C, Walker B. Catastrophic shifts in ecosystems. Nature. 2001;413 6856: 591–596.
  3. 3. Jackson JBC. Ecological extinction and evolution in the brave new ocean. Proc Nat Acad Sci. 2008;105: Supplement 11458–11465.
  4. 4. Smith JE, Shaw M, Edwards RA, Obura D, Pantos O, Sala E et al. Indirect effects of algae on coral: algae‐mediated, microbe‐induced coral mortality. Ecol Lett. 2006;9(7): 835–845. pmid:16796574
  5. 5. Pandolfi JM, Jackson JBC, Baron N, Bradbury RH, Guzman HM, Hughes TP et al. Are US coral reefs on the slippery slope to slime? Science. 2005;307(5716): 1725–1726. pmid:15774744
  6. 6. Vargas-Angel B,White D, Storlazzi C, Callender T, Maurin P. Baseline assessments for coral reef community structure and demographics on West Maui National Oceanic and Atmospheric Administration; 2017.
  7. 7. Sueur J, Pavoine S, Hamerlynck O, Duvail S. Rapid acoustic survey for biodiversity appraisal. PLoS ONE. 2008;3(12): e4065. pmid:19115006
  8. 8. Kennedy EV, Holderied MW, Mair JM, Guzman HM, Simpson SD. Spatial patterns in reef-generated noise relate to habitats and communities: evidence from a Panamanian case study. J Exp Mar Biol Ecol. 2010;395(1): 85–92.
  9. 9. Freeman SE, Rohwer FL, D’Spain GL, Friedlander AM, Gregg AK, Sandin SA et al. The origins of ambient biological sound from coral reef ecosystems in the Line Islands archipelago. J Acoust Soc Am. 2014;135: 1775–1788. pmid:25234977
  10. 10. Freeman LA, Freeman SE. Rapidly obtained ecological indicators in coral reef soundscapes. Mar Ecol Prog Ser.2016;501: 69–82.
  11. 11. Radford CA, Tindle CT, Montgomery JC, Jeffs AG. Modeling a reef as an extended sound source increases the predicted range at which reef noise may be heard by fish larvae. Mar Ecol Prog Ser. 2011;428; 167–174.
  12. 12. Akamatsu T, Okumura T, Novarini N Yan HY. Empirical refinements applicable to the recording of fish sounds in small tanks. J Acoust Soc Am. 2002;112(6); 3073–3082. pmid:12509030
  13. 13. Kraines S, Suzuki Y, Yamada K Komiyama H. Separating biological and physical changes in dissolved Oxygen concentration in a coral reef. Limnology and Oceanography. 1996;41(8): 1790–1799.
  14. 14. Minneart M. XVI On musical air-bubbles and the sounds of running water. The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science 1933;16(104): 235–248.
  15. 15. Leighton TG. The acoustic bubble ISBN: 0-12-44190-8; 1994.
  16. 16. Wilson CJ, Wilson PS Dunton KH. An acoustic investigation of seagrass photosynthesis. Marine Biology. 2012;159(10); 2311–2322.
  17. 17. Medwin H. In situ acoustic measurements of bubble populations in coastal ocean waters. J Geophys Res.1970;75(3): 599–611.
  18. 18. Tyree MT, Sperry JS. Vulnerability of xylem to cavitation and embolism. Ann Rev Plant Biol. 1989;40; 19–38.
  19. 19. Deane GB, Stokes MD. Scale dependence of bubble creation mechanisms in breaking waves. Nature. 2002;418(6900); 839–844. pmid:12192401
  20. 20. Wilson PS, Dunton KH. Laboratory investigation of the acoustic response of seagrass tissue in the frequency band 05–25 kHz. J Acoust Soc Am. 2009;125(4): 1951–1959. pmid:19354371
  21. 21. Ainslie MA, Leighton TG. Review of scattering and extinction cross-sections, damping factors, and resonance frequencies of a spherical gas bubble. J Acoust Soc Am. 2011;130(5): 3184–3208. pmid:22087992
  22. 22. Keeling RF. On the role of large bubbles in air-sea gas exchange and supersaturation in the ocean. J Mar Res. 1993;51(2): 237–271.
  23. 23. Freeman SE, Buckingham MJ, Freeman LA, Lammers MO, D’Spain GL. Cross-correlation, triangulation, and curved-wavefront focusing of coral reef sound using a bi-linear hydrophone array. The Journal of the Acoustical Society of America. 2015 Jan;137(1):30–41. pmid:25618036
  24. 24. Versluis M, Schmitz B, von der Heydt A, Lohse D. How snapping shrimp snap: through cavitating bubbles. Science. 2000 Sep 22;289(5487):2114–7. pmid:11000111
  25. 25. Au WW, Banks K. The acoustics of the snapping shrimp Synalpheus parneomeris in Kaneohe Bay. J Acoust Soc Am. 1998;103: 41–47.
  26. 26. Bohnenstiehl DR, Lillis A, Eggleston DB. The curious acoustic behavior of estuarine snapping shrimp: Temporal patterns of snapping shrimp sound in sub-tidal oyster reef habitat. PLoS ONE 11(1): e0143691. pmid:26761645
  27. 27. Lillis A, Eggleston DB, Bohnenstiehl DR. Estuarine soundscapes: distinct acoustic characteristics of oyster reefs compared to soft-bottom habitats. Marine Ecology Progress Series. 2014 May 28;505:1–7.
  28. 28. Staaterman E, Rice AN, Mann DA, Paris CB. Soundscapes from a Tropical Eastern Pacific reef and a Caribbean Sea reef. Coral Reefs. 2013 Jun 1;32(2):553–7.
  29. 29. Staaterman E, Paris CB, DeFerrari HA, Mann DA, Rice AN, D’Alessandro EK. Celestial patterns in marine soundscapes. Marine Ecology Progress Series. 2014 Aug 4;508:17–32.
  30. 30. Radford C, Jeffs A, Tindle C, Montgomery JC. Resonating sea urchin skeletons create coastal choruses. Mar Ecol Prog Ser. 2008;362: 37–43.
  31. 31. McCauley RD, Day RD, Swadling KM, Fitzgibbon QP, Watson RA, Semens JM. Widely used marine seismic survey air gun operations negatively impact zooplankton. Nature Ecology and Evolution. 2017;1(7); s41559–017.
  32. 32. Luczkovich JJ, Mann DA, Rountree RA. Passive acoustics as a tool in fisheries science. Transactions of the American Fisheries Society. 2008;137: 533–541.
  33. 33. Giorli G, Au WW, Neuheimer A. Differences in foraging activity of deep sea diving odontocetes in the Ligurian Sea as determined by passive acoustic recorders. Deep Sea Research Part I: Oceanographic Research Papers. 2016;107: 1–8.
  34. 34. Chapman NR, Price A. Low frequency deep ocean ambient noise trend in the Northeast Pacific Ocean. J Acoust Soc Am. 2011;129(5): 161–165.
  35. 35. Dahl PH, Miller JH, Cato DH, Andrew RK. Underwater ambient noise. Acoustics Today. 2007;3(1): 23–33.
  36. 36. Dunlop RA, Cato DH, Noad MJ. Your attention please: increasing ambient noise levels elicits a change in communication behaviour in humpback whales (Megaptera novaeangliae). Proceedings of the Royal Society of London B: Biological Sciences. 2010; prspb20092319.
  37. 37. Giorli G, Au WW, Ou H, Jarvis S, Morrissey R, Moretti D. Acoustic detection of biosonar activity of deep diving odontocetes at Josephine Seamount High Seas Marine Protected Area. J Acoust Soc Am. 2015;137(5): 2495–2501. pmid:25994682
  38. 38. Boyd JW, Varley J. The uses of passive measurement of acoustic emissions from chemical engineering processes. Chem Eng Sci. 2001;56(5); 1749–1767.
  39. 39. Preskitt LV, Vroom PS, Smith CM. A rapid ecological assessment (REA) quantitative survey method for benthic algae using photoquadrats with scuba. Pacific Science. 2004;58: 201–209.
  40. 40. Yentch CS, Yentsch CM, Cullen JJ, Lapointe B, Phinney DA, Yentsch SW. Sunlight and water transparency: cornerstones in coral research. J Exp Mar Biol Ecol. 2002;268(2): 171–183.
  41. 41. Sensors for sondes and monitoring. 2017;11:20. Available from: https://wwwwaterprobescom/sensors-for-sondes-and-monitoring.


Tau accumulations in the brains of woodpeckers


Woodpeckers experience forces up to 1200–1400 g while pecking. It is assumed due to evolutionary adaptations, the woodpecker is immune to brain injury. This assumption has led to the use of the woodpecker as a model in the development of sports safety equipment such as football helmets. However, it is unknown at this time if the woodpecker brain develops neuro-trauma in relation to the high g-forces experienced during pecking. The brains of 10 ethanol preserved woodpeckers and 5 ethanol preserved red-winged black bird experimental controls were examined using Gallyas silver stain and anti-phospho-tau. The results demonstrated perivascular and white matter tract silver-positive deposits in eight out of the 10 woodpecker brains. The tau positive accumulations were seen in white matter tracts in 2 of the 3 woodpeckers examined. No staining was identified in control birds. The negative staining of controls birds contrasted with the diffuse positive staining woodpecker sections suggest the possibility that pecking may induce the accumulation of tau in the woodpecker brain. Further research is needed to better understand the relationship.


In the central nervous system of humans, the protein tau assists in the assembly and stabilization of neuronal microtubules. Accumulations of tau can be seen in conditions ranging from normal aging to various neurodegenerative diseases such as Alzheimer’s disease [1]. In the disease state, through a process yet to be understood, tau dissociates from the axons and become hyperphosphorylated to form insoluble neurofibrillary tangles [2]. It is not entirely clear if these aggregates are responsible for the symptoms associated with neurodegenerative diseases.

Tau accumulations have also been observed in association with chronic traumatic encephalopathy (CTE) [3]. CTE is theorized to be the end result of repetitive mild traumatic brain (mTBI) injury. Repeated concussions has been suggested as a cause of CTE in contact sport athletes.

Recently, the CTE center at Boston University reported CTE in brain tissue of 110 out of 111 former National Football League (NFL) players studied [4]. However, CTE is not unique to football and has been identified in the brains of athletes who play soccer, rugby and hockey [5].

The prevalence of CTE and relationship between mTBI and the subsequent development of CTE has yet to be fully established. Currently, the disease can only be diagnosed by post mortem analysis utilizing immunohistochemistry with antibodies directed against p-tau.

The most prevalent pathological changes thought to be diagnostic of CTE are focal accumulation of abnormal hyperphosphorylated tau (p-tau) within neurons and astroglia distributed around small blood vessels located at the depths of cortical sulci [6]. Other staining patterns considered to be supportive of CTE include pre tangles and neurofibrillary tangles in the superficial layers of the cortex and the hippocampus; neuronal and astrocytic aggregates in subcortical nuclei; and ‘thread-like’ and ‘dot-like’ axonal tau staining patterns [5, 7, 8]

Because of the theorized association between mTBI and CTE, the prevention of TBI in athletics has become an important area of research. Due to it’s assumed resistance to neurotrauma, the woodpecker has become a model for the development of safety equipment such as football helmets and neck collars [9, 10].

The Picidae family of birds, which includes woodpeckers and sapsuckers, have several evolutionary anatomic adaptations to theorized to mitigate the enormous forces they experience while pecking. These include sharp beaks with upper and lower components which can move independently of each other while pecking and a long tongue that is capable of bracing the skull and brain during impacts. Other proposed protective adaptations include: thick neck muscles to dissipate force, and unique bony features of the skull [6]. It is assumed these evolutionary adaptations prevent neurotrauma in the woodpecker brain.

The majority of research using the woodpecker as a model for the development of safety equipment has focused on the biophysics of the woodpecker’s head. To date, only one paper has examined histologic sections of woodpecker brains in an effort to investigate the possible existence of brain injury in these birds. [11] This previous study utilized a modified Prussian blue stain (ferrocyanide) on the brains of two woodpeckers. Without describing their histologic findings, the authors concluded woodpeckers do not experience brain injury. Since its publication in 1976, this study has been sited in over 100 journal articles supporting the conclusion that woodpeckers do not incur brain injury in association with pecking behavior.

Despite the wide use of the 1976 paper, the neurobiological response of woodpeckers to repetitive head acceleration and deceleration is vastly unexplored. At this time, it is not known with any certainty if the brains of these animals experience neurotrauma in association with their pecking behavior. If so, the woodpecker may serve as an important animal model for the further study of CTE.

With the woodpecker model increasing in popularity as a potential source of protective equipment technology development, an in-depth and comprehensive study into woodpecker brain injury is warranted. [9] Given this, our group our group set out to determine if tau accumulations exist in the brains of woodpeckers.

Materials and methods

Protocol doi: 10.17605/

Avian specimens and brain tissue collection

All the bird specimens used in this project were generously donated from museum collections (see acknowledgments). The woodpeckers (n = 10) studied were of various species (Table 1) To extract the brains, the feathers and skin covering the skull were removed. The skull cap was cut with a rotary tool from just above the orbits to just below the occiput in a circular fashion to remove the skull from the brain tissue below it. The plate of bone that protects the optic center of the brain, which is somewhat analogous to the human’s tentorium cerebelli, was removed using forceps. The brain was gently pried away from the the skull, taking great care around the frontal lobes and brainstem areas. Finally, an 11 blade scalpel was used to sever the brain stem right below the level of the pons. The extracted brains were then placed in 70% ethanol until tissue processing could occur.

Histological and immunohistochemical procedures

The woodpecker and red winged black bird brains were cut into gross tissue sections according to anatomic landmarks. Tissue processing was performed the same for all tissue samples according to standard paraffin-embedding processing procedures [12].

The Gallyas silver stain was performed in accordance to a previous published methodology with some minor alterations at room temperature with slight agitation [13]. Experimental and control slides were sectioned to 15μm onto slides, deparaffinized, and rehydrated back to water. The slides were then placed in a 0.25% potassium permanganate for 15 minutes. Then, slides were incubated in 2% oxalic acid for 5 minutes. Following the oxalic acid incubation, the slides were rinsed in dH2O for 5 minutes before being placed in a 0.4% lanthanum nitrate/2% sodium acetate blocking solution for one hour. After the blocking step, the slides were rinsed again in dH2O for one minute before going into a 4% sodium hydroxide/10% potassium iodide/0.035% silver nitrate (added in that order) solution for four minutes. The slides were immediately placed into a 0.5% acetic acid for three, one minute rinses before being placed into developer. The slides were then put into 1% acetic acid for three minutes and then placed into dH2O. Mayer’s hematoxylin counterstain (Scytek, #HAQ999) was used by placing the silver stained sections into the stain for two minutes before a quick rinse in dH2O. The slides were differentiated with 0.1% sodium bicarbonate bluing agent until desired color was achieved. Slides were then dehydrated using the standard ethanol gradient and Histoclear before being coverslipped.

The immunohistochemical staining was achieved using a hybrid of previously published techniques. [14] The tissue was cut at 25μm and each section was placed into a 2cm in diameter steel-wire mesh container. The tissue slices were deparaffinized and rehydrated to water before entering the antigen retrieval step. Antigen retrieval was done by submerging the mesh containers into filtered 1x Tris/EDTA buffer pH 9 with 0.05% Tween 20 at 90°C for 20 minutes. After the 20-minute incubation, the mesh containers were then placed into filtered 1x TBS buffer pH 7.4 with 0.025% Triton X-100 and rinsed twice for five minutes each. After the second rinse, the mesh containers were removed and incubated in filtered 10% goat serum/1x TBS buffer pH 7.4 block for two hours. After incubation, tissue sections were removed from the mesh containers and placed into individual sterilized petri dishes containing antibody in sterile and filtered 1% goat serum 1x TBS buffer pH 7.4 overnight at 4°C with gentle agitation. The antibodies used were anti-phospho-tau S262 rabbit polyclonal (5μg/mL; Abcam, ab64193; Cambridge, MA) and anti-glial fibrillary associated protein rabbit polyclonal (5μg/mL; Bioss, bs-0199R; Woburn, MA). Following overnight incubation, the sections were removed from the primary antibody and placed back into the mesh containers to be rinsed with 1x TBS with 0.025% Triton X-100 twice, each for five minutes. The mesh containers were then placed into 0.3% sodium hydroxide 1x TBS buffer pH 7.4 for 15 minutes. After the sodium hydroxide incubation, the mesh containers were once again rinsed for three minutes in 1x TBS buffer pH 7.4 with 0.025% Triton X-100. The sections within the mesh containers were removed once again, and placed into small, sterile petri dishes containing horseradish peroxidase secondary antibody (1μg/mL; Abcam, ab6721; Cambridge, MA) in 1x TBS buffer pH 7.4 for one hour at room temperature. After the hour incubation, the sections were placed back into their mesh containers and rinsed twice, for five minutes each, in 1x TBS buffer pH 7.4. The sections were then ready for the 3,3′-diaminobenzidine (DAB) reaction. The DAB chromagen (Biocare Medical; DB801R; Concord, CA) reaction was carried out under a microscope until adequate staining was achieved. Sections were then immediately placed back into their mesh containers and rinsed with dH2O for five minutes. Following the dH2O rinse, the sections were counterstained with Mayer’s hematoxylin counterstain (Scytek, #HAQ999) for two minutes before a quick rinse in dH2O. The sections were then differentiated with 0.1% sodium bicarbonate bluing agent until desired color was achieved. After another quick rinse in dH2O, the sections were dehydrated using the standard ethanol gradient and Histoclear while in the mesh containers. After the last Histoclear incubation, the sections were removed from their mesh containers and free-floated into a large crystallization dish containing Histoclear. The sections were coaxed onto clean glass slides with fine tipped forceps. The slides were then cover slipped with mounting medium and placed onto a 37°C warming plate overnight.


Gallyas silver stain

Because Gallyas stain has a high degree of sensitivity and specificity for neurofibrillary tangles and axonal injury, it was utilized to detect the presence of neuronal and/or white matter tract damage throughout the entire woodpecker brain.

A section of human cortex with confirmed Alzheimer’s disease was used as a positive staining control while red-winged black birds’ brains (n = 5) were used as experimental controls for all staining methods.

Positive silver accumulations were identified in 8 out of the 10 woodpeckers studied. Several patterns of positive Gallyas staining were identified in the woodpecker population including: focal perivascular deposits, which were mostly subpial, (Fig 1A, 1C and 1D); focal whole astrocyte staining (Fig 1B); dot-like staining within axonal tracks; and wide-spread thread-like axonal track staining in deep white matter tracks. The majority of the observed silver-positive staining patterns were identified in the frontal pole of the brain. Very rare staining was detected in the occipital region, and no staining was seen in the cerebellum.


Fig 1. Perivascular and axonal track pathology of Dryocopus lineatus and picoides pubescens.

Perivascular Gallyas silver positive pathology in the cortex of the frontal lobe (A). Damaged axonal tracts (B) with axonal swellings (arrow) in the subcortical white matter of the frontal lobe. Subpial perivascular staining (C) of the frontal lobe. Perivascular silver positive staining in (D) in a superficial region of the frontal cortex. Experimental controls (E).

Focal perivascular silver-positive deposits were found in 40% of the woodpeckers. The most abundant positive silver staining pattern was “thread-like” axonal white matter tracts of the frontal lobe, which appeared in 80% (n = 10) of the woodpeckers studied (Fig 2).


Fig 2. Axonal tract pathology of Dryocopus lineatus and Picoides pubescens.

Gallyas silver positive axonal tract staining in the corpus callosum (A) and the mediolateral central gray area of the midbrain (B and C) in the woodpecker brain (A) [15].

No positive Gallyas staining was observed in the control birds (n = 10).


Following the analysis of the Gallyas silver stain, we proceed to immunohistochemical verification of the suspected tau accumulations throughout the entire woodpecker brain. Samples were stained for phosphorylated tau (S262).

Due to the poor preservation of the woodpecker tissue, successful immunohistochemistry was accomplished in only 3 birds; the remainder of the tissue samples degraded during processing. Attempts to alter the immunohistochemistry methodology proved to not help with the degradation of the tissue samples. Immunohistochemistry was performed on all control birds (n = 5).

In the woodpecker specimens where tau immunostaining was possible, tau-positive accumulations were identified in the same regions highlighted originally by the Gallyas silver stain. The morphology of astrocytes identified by GFAP were used to determine some of the cells staining with tau were in fact astrocytes. This assisted in confirming the silver-positive accumulations identified were in fact comprised of tau protein. Specifically, tau immunostaining demonstrated perivascular deposits, whole astrocyte staining, and ‘thread-like’ axonal track staining in 2 of the 3 woodpecker brains evaluated (Fig 3).


Fig 3. Anti-phospho-tau immunostaining in Dryocopus lineatus.

Tau-positivity in the midbrain (A and B) and the corpus callosum (C) of the Dryocopus lineatus brain. The axonal tract staining demonstrates a thread-like pattern, similar to that seen with Gallyas sliver staining (Fig 2). Occasional intracellular tau-accumulations were identified within neurons (D).

No positive immunostaining was identified in any of the control birds.

Tau immunohistochemistry was successfully completed in three woodpecker brains. Two of the three woodpecker brains demonstrated wide-spread thread-like axonal track staining with tau. We observed positive GFAP-staining in these same two birds (Fig 4). One woodpecker failed to demonstrate positive sliver and tau staining. This same bird also showed negative GFAP immunohistostaining. Interpreted collectively, this suggests that the tau accumulations we identified are pathological.


Fig 4. Anti-GFAP immunostaining in Dryocopus lineatus.

Immunostaining with GFAP demonstrated rare GFAP-positive astrocytes. Astrocyte morphology included thorn-shaped (A), typical star-like (B) and tufted (C). Tau immunohistochemistry demonstrated rare tau accumulations in cells morphologically consistent with astrocytes within the grey matter (D, arrow).

In summary, focal subpial perivascular deposits; focal whole astrocyte staining; dot-like staining within axonal tracks; and wide-spread thread-like axonal track staining in white matter tracks were identified in the frontal poles of 80 percent of the study population. Strikingly, no staining was identified in the control bird population.

These observations were confirmed by a board-certified neuropathologist.


To date, there have been no histologic studies exploring the potential existence of neurotrauma in woodpeckers. While it is unknown if the forces associated with pecking behavior could result in traumatic brain injury, it is interesting that the majority of the woodpecker specimens in our study displayed focal silver-positive deposits, some of which were confirmed to be tau by immunohistochemistry, while no staining was observed in the control birds.

The anatomic locations and staining patterns of the lesions identified in the brains of woodpeckers shares some similarities to human CTE. In humans, CTE is most prominent in the frontal and temporal lobes of the brain with a spectrum of tau deposition patterns including focal perivascular staining, astrocytic inclusions, ‘thread-like’ and ‘dot-like’ axonal staining patterns (5, 7). In the woodpecker, we identified similar focal perivascular staining, astrocytic inclusions, ‘thread-like’ and ‘dot-like’ axonal staining patterns which were confined to the frontal lobe of the brain. The woodpecker brain lacks the gyri and sulci seen in the human brain. Because of this, we could not evaluate for lesions located at the depth of sulci, as seen in human CTE.

The prominent frontal and temporal anatomic locations of CTE lesions are thought to be due to the distribution of force experienced in head collision [14]. In the woodpecker, much of the force of pecking is thought to be dissipated through the frontal regions of the skull and brain, as well. Therefore, it is not surprising to see potential areas of injury limited to this anatomic location in woodpecker brain.

In the human brain, tau accumulations are also known to occur as part of the normal aging process. Though it cannot be entirely ruled out, age-related tau accumulation in the woodpecker is an unlikely explanation for our findings. Our study population included one juvenile woodpecker (Sphyrapicus varius) which demonstrated the full spectrum of tau accumulations observed in the majority of the adult population suggesting the tau accumulations seen in our study might not be age-related. This notion is further supported by the lack of observable staining in the control population which was comprised of entirely adult birds.

Given the complete lack of staining in the control population and the unlikely scenario of age-related changes, our findings suggest there might be an association between repetitive pecking behavior and tau accumulations in the woodpecker population.

There are several limitations to this study. It cannot be concluded at this time the if the histologic changes identified in our study are the direct result of the repeated, high force pecking woodpeckers endure everyday. The limited number of woodpeckers (n = 10) and control birds (n = 5) utilized in this study are insufficient to established a correlation between pecking behavior and subsequent neurotrauma.

It is not known from our study whether the tau accumulations are pathological or result in behavioral changes in woodpeckers. However, our findings of silver and tau accumulations solely in pecking birds warrant further investigation into this possibility.

There are numerous anatomic differences between the skulls and brains of woodpeckers. It may be that the anatomic adaptations of the woodpecker produce stress in regions of the brain in different locations than humans. Further research is necessary to understand how our findings can be translated to the human population.

Due to the increased tau deposition in our woodpecker population, the brains of woodpeckers are an important area for future research. Further studies are needed to determine what iso-forms of tau are being deposited in the woodpecker brain and if these deposits are pathological. The continued study of the response of the woodpeckers’ brain to pecking is necessary to assure current head protection technology based on the woodpecker model is providing adequate protection in athletes. Our findings also suggest the woodpecker may be a suitable animal model for the further study of CTE.

Worldwide trends in volume and quality of published protocols of randomized controlled trials



Publishing protocols of randomized controlled trials (RCT) facilitates a more detailed description of study rational, design, and related ethical and safety issues, which should promote transparency. Little is known about how the practice of publishing protocols developed over time. Therefore, this study describes the worldwide trends in volume and methodological quality of published RCT protocols.


A systematic search was performed in PubMed and EMBASE, identifying RCT protocols published over a decade from 1 September 2001. Data were extracted on quality characteristics of RCT protocols. The primary outcome, methodological quality, was assessed by individual methodological characteristics (adequate generation of allocation, concealment of allocation and intention-to-treat analysis). A comparison was made by publication period (First, September 2001- December 2004; Second, January 2005-May 2008; Third, June 2008-September 2011), geographical region and medical specialty.


The number of published RCT protocols increased from 69 in the first, to 390 in the third period (p<0.0001). Internal medicine and paediatrics were the most common specialty topics. Whereas most published RCT protocols in the first period originated from North America (n = 30, 44%), in the second and third period this was Europe (respectively, n = 65, 47% and n = 190, 48%, p = 0.02). Quality of RCT protocols was higher in Europe and Australasia, compared to North America (OR = 0.63, CI = 0.40–0.99, p = 0.04). Adequate generation of allocation improved with time (44%, 58%, 67%, p = 0.001), as did concealment of allocation (38%, 53%, 55%, p = 0.03). Surgical protocols had the highest quality among the three specialty topics used in this study (OR = 1.94, CI = 1.09–3.45, p = 0.02).


Publishing RCT protocols has become popular, with a five-fold increase in the past decade. The quality of published RCT protocols also improved, although variation between geographical regions and across medical specialties was seen. This emphasizes the importance of international standards of comprehensive training in RCT methodology.