1 Supplementary materials 2 The method descrided here are described in detail by Li et al. (2015), for more 3 detailed information, please refer to their publication. 4 Estimation of evaporation rate and effective diffusion coefficient in biofilm 5 The porosity of the biofilm (θ) was estimated from the volume of cells (biovolume) in 6 the biofilm divided by the total biofilm volume. The biovolume was calculated as 7 described by Hillebrand et al. (1999) from suspended biofilm samples with a 50 μm 8 resolution by means of a custom-made hand microtome (Fig. S2). The biofilm 9 immobilized on a polycarbonate membrane was placed onto a height-adjustable stage 10 on a thin layer of culture medium. The stage allows levelling the biofilm sample into 11 the sectioning plane by means of a micrometer screw. The sectioning plane was 12 defined by a solid brass support over which a rigid microtome blade was passed 13 manually to remove the overlying algal biofilm (Fig. S2). 14 Evaporation rate during the measurement was 0.68 mL min-1 (monitored by recording 15 the change of the volume in the medium container, Fig S1), which corresponded to a 16 convective flow rate of 5.4 µm s-1 inside the biofilm, which was calculated as: 18 π’ = ππ /(π΄π β π) 17 19 (Eq.S1), qe is the rate of liquid volume lost, Ae is the exposed surface area of the biofilm. 1 20 Microsensor setup 21 A schematic representation of the experimental setup for microsensor measurements 22 based on previous works by Gieseke and de Beer (2004) is given in Fig. S1: the 23 biofilm immobilized on a polycarbonate membrane was placed onto a moist glass 24 fiber inside the biofilm measurement cell under a controlled atmosphere of 25 compressed air at a flow rate of 1 L min-1 (Fig. S1A). The non-inoculated area of the 26 polycarbonate filter and glass fiber were covered with a black plastic foil to exclude 27 light and gas exchange in this area. 50 mL of BBM culture medium were circulated 28 though the measurement cell with a flow rate of 3 mL min-1 by means of a peristaltic 29 pump and were exchanged every 1 hour during the measurement. Light was supplied 30 from the front side by a halogen lamp (KL-1500, Schott, Mainz, Germany) equipped 31 with a 3-fold splitter to ensure even distribution. The movement of the sensor was 32 enabled by a computer-controlled micromanipulator (Pollux Drive, PI miCos, 33 Eschbach, Germany). Microsensor signals were amplified and converted into digital 34 data (DAQpad 6015 and 6009, National Instruments, Munich, Germany) prior to their 35 further processing on a computer (Fig. S1B). Software used for system control and 36 data acquisition was custom made (can be acquired upon request from Dr. Lubos 37 Polerecky, Utrecht University, Netherlands, l.polerecky@uu.nl). Light-dark shift 38 measurements (Revsbech, 1989) were carried out, and the data acquired was 39 processed as described in the following section. Linear calibration of oxygen 40 microsensor was carried out as described by Revsbech (1983) using BBM medium 2 41 saturated with nitrogen gas or compressed air. A shutter connected to a timer was 42 used to control light-dark cycling: the total dark period was 3 s, and a linear 43 regression slope of oxygen concentration change between 0.7 and 2.1 s of the dark 44 period was determined as the measured value. For each measurement depth, three 45 light-dark cycles were performed after steady state was reached (this could be as long 46 as 10 min), and these 3 values acquired were considered as triplicates. The PAR light 47 intensity of a fixed depth was determined as the average value of a 3 s measurement 48 period. 49 Gross photosynthetic productivity calculation 50 Take the derivative of time on both sides of the dissolved oxygen mass transfer 51 equation, and assume the time t and spatial variables x are not dependent (as during 52 the measurement period, the thickness of the biofilm hardly changes) (Bergman et al., 53 2011, Revsbech and Jorgensen, 1983): 55 ππΆ ππΆ π2 ( ) π( ) ππΆ ππ‘ + π’ ππ‘ − ππ /ππ‘ π( )/ππ‘ = 0/ππ‘ − ππ π /ππ‘ + π·π π ππ‘ ππ₯ 2 ππ₯ 54 (Eq.S2). 56 The term ∂Rs/∂t represents the rate change of the removal of dissolved gaseous 57 species at the submerged biofilm surface due to the flow of bulk liquid, or, in this 58 case, of the gas phase above the biofilm: 59 ππ π /ππ‘ = (π·π (πΆ − πΆπ ) (πΆ − πΆπ ) + π’ )/ππ‘ π₯2 π₯ 3 =( 61 π·π π’ + ) β ππΆ/ππ‘ π₯2 π₯ 60 (Eq.S3). 62 Cs represents the oxygen concentration on the very surface of the aerial biofilm, 63 which is in equilibrium with the phase above and is considered to be a constant. 64 Considering that the rate of respiration remains stable during the measurement period 65 (Glud et al., 1992), and take ∂C/∂t=P(x, t), as ∂C/∂t is a function of depth x and time t. 66 Substitute Eq.S3 into ∂Rs/∂t and ∂C/∂t to P(x, t), Eq.S2 becomes: 68 π 2 (π(π₯, π‘)) π(π(π₯, π‘)) ππ(π₯, π‘) π·π π’ = 0 − 0 + π·π +π’ − ( 2 + ) β π(π₯, π‘) 2 ππ‘ ππ₯ ππ₯ π₯ π₯ 67 (Eq.S4). 69 Eq.S4 is a nonhomogeneous diffusion equation with a sink and can be solved by using 70 any of the methods summarized by Crank or Polyanin (Crank, 1975; Polyanin, 2002). 71 Using an initial condition P(x, t = 0) = P(x, 0) and a boundary condition ∂P(x = 0, 72 t)/∂t = 0, the exact solution of Eq.S4 at t = τ can be expressed as: πΏ 74 π(π₯, π) = ∫ π(π¦, 0) β [π −(π¦−π₯+π’βπ) ( ) 4βπ·π βπ 0 −π −(π¦+π’βπ) ( ) 4βπ·π βπ β ππππ( π₯ √4 β π·π β π )] β ππ¦ 73 (Eq.S5), 75 δ is the total depth of the biofilm, or, alternatively, the depth until which the shift from 76 light to dark has an effect on the dissolved oxygen concentration. P(x, τ) is the 77 measured value from the light-dark shift method with measurement time τ at 78 measurement depth x, and P(y, 0) is the actual photosynthetic activity at position y 4 79 (the integrating variable), which is the desired term. For measurements taken at all 80 depths, the measured value P(x, τ) at a depth x depends on the P(y, 0) across the 81 complete photosynthetically active region of the biofilm (surface to depth δ). 82 Eq.S5 is a Fredholm integral equation of the first kind (Kress, 2014), and its 83 approximated solution can be calculated using Tikhonov regularization (Tikhonov et 84 al., 1995) with a non-negative constrain applying the active-set algorithm purposed by 85 Gill et al. (1981): Discrete δ by n, and let P(xj, τ) = gj, P(yi, 0) = hi; (i, j = 1, 2, …, n) 86 and write Eq.S5 in the discrete operator notion. π ππ = ∑ ππ,π β βπ 88 π 87 (Eq.S6), 89 ki,j is the operator term which contains both the term that reflects the effect of 90 diffusion and advection of dissolved oxygen inside the biofilm (inner term, first term 91 in square bracket in Eq.S5) and the term that reflects the effect of dissolved oxygen 92 removal due to surface flow (surface term, second term in square bracket in Eq.S5). In 93 matrix notion, let G be a vector containing gj, H a vector containing hi, and K a matrix 94 containing the operator terms ki,j: πΊ =πΎβπ» 96 95 97 (Eq. S7). Rewrite Eq.S7 as: 5 (πΊππ + πΊπ π’ππ ) = (πΎππ + πΎπ π’ππ ) β π» 99 98 (Eq.S9), 100 The subscripts in and surf represent the effect from the inner term and the surface 101 term in Eq.S5 respectively, operators Kin, Ksurf can be calculated using Eq.S5, as well 102 as gin and Gsurf, if an H is given. Apply the Tikhonov regularization, the problem in 103 Eq.11 thus becomes: 2 min {β(πΊππ + πΊπ π’ππ ) − (πΎππ + πΎπ π’ππ ) β π»β + π2 β βπΏ β π»β2 } 104 105 107 πΉ≥0 or, 2 min {β(πΊππ − πΎππ β π») + (πΊπ π’ππ − πΎπ π’ππ β π»)β + π2 β βπΏ β π»β2 } πΉ≥0 106 (Eq.S10). 108 Observing the surface term in Eq.S5, notice the maximum value of the surface term is 109 controlled by the position of the actual activity y, but a complement error function 110 dependent solely on the position of the measurement taken (x) controls the final value. 111 Furthermore, for a given G, under the same measurement conditions, Gsurf always has 112 the same set of values. Thus, for a given G, the minimization of the term 113 βπΊπ π’ππ − πΎπ π’ππ β π»β will always yield the same solution. As a result, the problem in 114 Eq. S10 can be simplified to: 116 2 min{βπΊππ − πΎππ β π»β2 + π2 β βπΏ β π»β2 } π≥0 115 (Eq. S11). 6 117 Eq.S11 does not contain the surface term, but the solution of the problem still leads to 118 H, which is the desired original profile. The L-curve method purposed by Hansen & 119 O'Leary (1993) was used to find the best regularization parameter λ. 120 The treatment procedure was coded in MATLAB (version 2013a, MathWorks, 121 Ismaning, Germany), and the inbuilt quadprog was used for solving the minimization 122 problems. L-curve function from the regularization tools (Hansen, 1994) was used to 123 find the λ value. The mean values of triplicates acquired from the microsensor 124 measurement were randomly added or subtracted with a random value in range of the 125 calculated standard deviation (SD) at the same position. These values were used as 126 input for the calculation. The calculation procedure was repeated 3 times, and the 127 mean value of the 3 results was taken as the final result (a SD can also be calculated). 128 Data interpolation, if needed, was done by simply connecting two measured data 129 points with a straight line (linear interpolation, using MATLAB inbuilt function 130 interp1). 7 131 References for supplementary materials: 132 Bergman TL, Lavine AS, Incropera FP, DeWitt DP. 2011. Fundamentals of heat and 133 mass transfer. 7th ed. Hoboken, New Jersey, USA: John Wiley & Sons. 1048 p. 134 Crank J, editor. 1975. Mathematics of diffusion. 2nd ed. Oxford, UK: Oxford 135 University Press. 414 p. 136 Gieseke A, de Beer D. 2004. Use of microelectrodes to measure in situ microbial 137 activities in biofilms, sediments, and microbial mats. In: Kowalchuk GA, de Bruijn 138 FJ, Head IM, Akkermans AD, van Elsas JD, editors. Molecular Microbial Ecology 139 Manual. The Netherlands: Springer. p 3483-3514. 140 Gill PE, Murray W, Wright MH. 1981. Practical optimization. Waltham, 141 Massachusetts, USA: Academic Press. 401 p. 142 Glud RN, Ramsing NB, Revsbech NP. 1992. Photosynthesis and photosynthesis- 143 coupled respiration in natural biofilms quantified with oxygen microsensors. Journal 144 of Phycology 28(1):51-60. 145 Hansen P. 1994. REGULARIZATION TOOLS: A Matlab package for analysis and 146 solution of discrete ill-posed problems. Numerical Algorithms 6(1):1-35. 8 147 Hansen PC, O’Leary DP. 1993. The Use of the L-Curve in the Regularization of 148 Discrete Ill-Posed Problems. SIAM Journal on Scientific Computing 14(6):1487- 149 1503. 150 Hillebrand H, Durselen CD, Kirschtel D, Pollingher U, Zohary T. 1999. Biovolume 151 calculation for pelagic and benthic microalgae. Journal of Phycology 35(2):403-424. 152 Kress R, editor. 2014. Linear integral equations. 3rd ed. New York City, USA: 153 Springer. 412 p. 154 Li T, Podola B, de Beer D, Melkonian M. 2015. A method to determine 155 photosynthetic activity from oxygen microsensor data in biofilms subjected to 156 evaporation. Journal of Microbiological Methods 117:100-107. 157 Polyanin AD. 2002. Handbook of linear partial differential equations for engineers 158 and scientists. Boca Raton, Florida: Chapman&Hall/CRC Press. 800 p. 159 Revsbech NP. 1983. In Situ Measurement of Oxygen Profiles of Sediments by use of 160 Oxygen Microelectrodes. In: Gnaiger E, Forstner H, editors. Polarographic Oxygen 161 Sensors. Berlin/Heidelberg, Germany: Springer. p 265-273. 9 162 Revsbech NP, Jorgensen BB. 1983. Photosynthesis of benthic microflora measured 163 with high spatial-resolution by the oxygen microprofile method - capabilities and 164 limitations of the method. Limnology and Oceanography 28(4):749-756. 165 Tikhonov AN, Goncharsky A, Stepanov VV, Yagola AG. 1995. Numerical methods 166 for the solution of ill-posed problems. Dordrecht, the Netherlands: Springer. 254 p. Supplementary materials figure captions 167 Figure S1 168 Experimental setup for microsensor studies in non-submerged microalgal biofilms. A: 169 Schematic drawing of the vertical cross section of the measurement in artificial 170 biofilms. Solid arrows in the glass fiber tissue indicate flow direction of the culture 171 medium. B: Schematic drawing of the complete experimental setup, dotted lines with 172 arrow indicate signal flow, whereas solid arrows represent medium and gas flows in 173 the system. 174 Figure S2 175 Schematic drawing of the experimental setup used for biofilm sectioning. The sample 176 holder allows moving the biofilm in vertical direction by means of a micrometer 177 screw (indicated by the thin solid arrows). The biomass above the sectioning plane 10 178 was removed with a solid microtome blade. The cutting direction is indicated by the 179 thick arrow. 11