Grain size modulates volcanic ash retention on crop foliage and potential yield loss

Ashfall from volcanic eruptions endangers crop production and food security, while 15 jeopardising agricultural livelihoods. As population in the vicinity of volcanoes continues to grow, strategies to reduce volcanic risks to and impacts on crops are increasingly needed. Current models of crop vulnerability to ash are limited. They also rely solely on ash thickness 18 (or loading) as the hazard intensity metric and fail to reproduce the complex interplay of other volcanic and non-volcanic factors that drive impact. Amongst these, ash retention on crop leaves affects photosynthesis and is ultimately responsible for widespread damage to crops. In 21 this context, we carried out greenhouse experiments to assess how ash grain size, leaf pubescence and humidity conditions at leaf surfaces influence the retention of ash (defined as the percentage of foliar cover coated with ash) in tomato and chilli pepper plants, two crop 24 types commonly grown in volcanic regions. For a fixed ash mass load (~570 g m -2 ), we found that ash retention decreases exponentially with increasing grain size and is enhanced when leaves are pubescent (such as in tomato) or their surfaces are wet. Assuming that leaf area 27 index ( LAI ) diminishes with ash retention in tomato and chilli pepper, we derived a new expression for predicting potential crop yield loss after an ashfall event. We suggest that the measurement of crop LAI in ash-affected areas may serve as an impact metric. Our study 30 demonstrates that quantitative insights into crop vulnerability can be gained rapidly from controlled experiments. We advocate this approach to broaden our understanding of ash-plant interaction and to validate the use of remote sensing methods for assessing crop damage and 33 recovery at various spatial and time scales after an eruption.


36
The livelihood and food security of hundreds of millions of people living near and on volcanoes intricately depend on agriculture (Small and Naumann, 2001;Brown et al., 2015). However, farming activities in these regions are exposed to short-term, i.e. usually less than one year, 39 negative impacts of volcanic eruptions, an issue amplified by the expanding population living under volcanic risk (Brown et al., 2015;Freire et al., 2019). Where cropping activity dominates (for example, in Indonesia), widespread damage to agriculture during eruptive activity arises 42 from crop exposure to ashfall (e.g. Burket et al., 1980;de Guzman, 2005;Tampubolon et al., 2018), causing adverse effects that range from temporary perturbations in leaf physiology to irreversible mechanical damage (Eggler, 1948;Blong, 1984;Grishin et al., 1996;Ayris and 45 Delmelle, 2012). As a result, crop fields impacted by ash deposition produce lower or poorquality harvests that can translate into significant economic losses to farmers and food shortages at the local or even regional scale, and even more so when subsistence agriculture dominates 48 (Neild et al., 1998;Ligot et al., 2022).
In this context, the development of strategies that can support disaster risk reduction and strengthen resilience for agrarian communities in volcanically active regions is critical, 51 especially in less-economically developed countries (FAO, 2021). Such measures require a sound understanding of agriculture vulnerability to ashfall (UNDRO,1980;Jenkins et al., 2015;Craig et al., 2021). Over the past 15 years, a dozen or so of post-eruption impact assessments 54 (post-EIA) have contributed to document the responses of farming systems exposed to ash (e.g. Wilson et al., 2011;Magill et al., 2013;Blake et al., 2015;Craig et al., 2016a;Craig et al., 2016b;Ligot et al., 2022). These field-based investigations have 57 underpinned the development of empirical relationships that link ash accumulation (also referred to as ash mass load or deposit thickness) to an estimated level of production loss for representing accumulations encountered at distal sites (and over wide areas) affected by ash fallout from explosive eruptions (Fierstein and Nathenson, 1992;Jenkins et al., 2022). Pretests carried out with higher ash loads (≥ 1000 g m -2 ) already led to lodging of some tomato 153 and chilli pepper plant specimens, a phenomenon that needed to be avoided in order to maximise the experiment's reproducibility. Neild et al. (1998) and Craig (2015) consider that an ash mass load of 6-30 kg m -2 on plants leads to mechanical damage. Our observations 156 indicate that lower loads can affect crop plants. In other words, the threshold value above which mechanical injury occurs varies with plant phenology (i.e. the combination of genotype and environment). 159 The selected ash load was applied uniformly to each plant using a homemade ashfall simulator (Fig. S3). The device consists of a 135 cm-high PVC tube (of diameter 29.5 cm) with three 1-mm opening meshes placed at 75, 110 and 120 cm from the tube base. The ash 162 fractions <1000 µm were poured carefully through a 2 cm-mesh sieve installed on the top of the PVC tube. Bouncing of the ash particles passing through the three inner 1-cm sieves allowed formation of a uniform deposit. Application of the coarsest ash (1000-2000 µm) was 165 carried out with the same device, but the inner meshes were removed. Wet conditions at leaf surfaces were obtained by spreading ~1.5 g of water on each plant using a commercial manual sprayer held one meter above the ground. In order to simulate the presence of water droplets 168 on plant leaves, we applied four sprays of water, one in each cardinal direction just before ash treatment. Water spraying of the plant foliage, ash application and photo acquisition all took place within the black chamber. Less than five minutes elapsed between the spraying 171 operation and photo acquisition of the ash-treated plant (Fig. S4).

Estimating the foliar cover from digital photos
We took photos of each plant before and immediately after ash treatment (Fig. S4). To 174 minimise uncontrolled variations in light colour and brightness, plants were photographed in a 1.6 x 1.2 x 2.2 m black chamber equipped with four led bulbs (6.5 W, cold white, Fig. S3 and   S4). We used a DX Nikon camera with an AF-S DX NIKKOR 18-55mm f/3.5-5.6G VR II 177 lens mounted on a 0.9 m-high tripod. Sheets of paper were placed on the floor and plant pot to produce a uniform background. A ribbon placed in a fixed position provided a reference scale.
We analysed the digital photos taken just before and after ash application with ImageJ 1.52 180 (Schindelin et al., 2015). The foliar cover, a measure of the vertical projection of exposed leaf area, was estimated using a dedicated macro (https://github.com/NoaLigot/ImageJ-macro.git).
While digital photos are recorded as a raster of red/green/blue (RGB) pixels, the values are not 183 standardised and can vary depending on the camera (Darge et al., 2019). The ImageJ macro transforms the RGB colour space into the International Commission on Illumination (CIE) 1976 L*a*b* colour space (McLaren, 1976), which has linear measures of lightness (L*) and 186 two colour dimensions (a* and b*). The a* dimension represents a spectrum from green (negative) to magenta (positive) and the b* dimension represents a spectrum from blue (negative) to yellow (positive). The a* attribute is useful to identify green pixels and was used 189 in the ImageJ macro to identify and select green parts of leaves. Values of 1 and 0 are attributed to a green and non-green (background) pixel, respectively. This allows delineation of the shape of the green leaf portion and calculation of its surface area.

Data treatment
The percentage of foliar cover coated with ash was inferred for each plant by comparing the foliar cover estimated from the image analysis, before and after ash application. A Tukey 195 HSD (Honest Significant Difference) test was applied to determine if means differ between treatments. Tomato and chilli pepper plant measurements carried out under dry and wet leaf surface conditions were processed separately, i.e. four sub-datasets were used in order to 198 compare the means separately for each combination of crops and moisture conditions.

Foliar cover coated with ash
The percentage of foliar cover coated with ash ranged from 0 to 99%, with an average value of 36 ± 33% (Table S1). The effect of ash grain size, humidity conditions at leaf surfaces and leaf pubescence on the foliar cover coated with ash is illustrated in Fig. 1. In general, foliar 204 cover coated with ash increased with decreasing ash grain size. Grain size ≥ 500 µm covered only 10% of the foliar cover, with coverage increasing up to ~90% for ash ≤ 90 µm. Wetting of tomato and chilli pepper leaves prior to ash application had no significant effect on the 207 retention of fine ash (≤ 90 µm). Nevertheless, significant higher tomato and chilli pepper leaf surface coverages (+17 ± 5% and +31 ± 10%) were inferred for intermediate ash grain sizes between 90 and 500 µm (Table S1, S2). We also note that for the ash grain size ranges 125-210 250 and 250-500 µm in dry conditions, coverage of tomato leaves with ash was significantly greater, by ~30 and 20% on average, compared to chilli pepper leaves. 213 Figure 1: Percentage of foliar cover coated with ash for tomato plant, i.e. which has pubescent leaves, (a) and chilli pepper plant, which has glabrous leaves (b). The percentage of foliage cover was measured for the six grain size ranges tested in dry and wet conditions at leaf 216 surfaces. Each boxplot represents 15 repetitions. The median value sits within the box and represents the centre of the data. Fifty % of the data values lie above the median and 50% lie below the median. Measurement outliers are displayed as dots.

Quantifying ash retention as a function of grain size
Using the experimental results obtained for tomato and chilli pepper (Fig. 1), we predicted the percentage of foliar cover coated with ash as a function of grain size, when leaf surfaces are 222 dry or wet. Five convex models (i.e. exponential decay, power curve, rectangular hyperbola, asymptotic curve and logarithmic curve) were fitted to the data points using the aomisc and nlme packages in R (Onofri, 2020;Pinheiro and Bates, 2022) (Fig. S5). The median grain size 225 was used to represent the corresponding grain size range. A lack-of-fit sum of squares test was applied to evaluate the relevance of each model. Since the five models have different numbers of parameters, their test statistics (F*) could not be compared directly. Instead, the 228 models were assessed based on their p-values (Table S3). All the models have p-values > 5%, with no evident lack-of-fit. The exponential decay model had the highest p-value for the four sub-datasets (0.82, 0.98, 1, 1 for dry tomato, wet tomato, dry chilli pepper and wet chilli 231 pepper, respectively) and it was chosen for the predictions.
Quantile regressions using the exponential decay model indicate that for 500 µm ash particles, there is a 50% chance to cover ~10 and ~27% of tomato foliar cover in dry and wet 234 conditions, respectively (Fig. 2). Similarly, for chilli pepper, foliar covers of <1 and 20% are estimated in dry and wet conditions, respectively. By the same tenet, there is a 50% probability that ash with a median of 63 μm in diameter covers up to ~67% (dry conditions) 237 and ~77% (wet conditions) of the foliar cover in tomato, and ~51% (dry conditions) and ~78% (wet conditions) of the foliar cover in chilli pepper.

Distribution of ash retention on the foliar cover
In addition to controlling ash retention on leaves, grain size, conditions of humidity at leaf surfaces and leaf pubescence affect the location of ash retention (Fig. 3). For tomato plants in 246 dry conditions, ash ≤ 90 µm tended to be lodged on the leaf surface wherever it had settled.
For glabrous chilli pepper leaves, leaf angle dictates if the ash particles remain on the leaf surface after deposition or slide off and relocate elsewhere. Ash with intermediate grain sizes 249 between 90 and 500 µm behaved differently, depending on humidity conditions. For both 13 tomato and chilli pepper plants, the ash material was found mainly along the primary and secondary veins of the horizontal upper leaves when they were dry. However, in wet 252 conditions, ash was more homogeneously distributed over the leaf surface. Coarser ash (≥ 500 µm) accumulated preferentially in the folds of growing leaves.

255
Figure 3: Photos processed with ImageJ of tomato and chilli pepper plants before (control) and after exposure to ~570 g m -2 of ash varying in grain size (≤ 90, 90-125, 125-250, 250-500, 258 500-1000, 1000-2000 µm) and in dry and wet conditions at leaf surfaces. The part of the foliar cover depicted in black corresponds to the green leaf surface area that was not covered with ash. The image surface area is equivalent to ~800 cm². The original photos of the ash-covered 261 plants are provided as supplementary material (Fig. S6).

Influence of grain size on ash retention 264
The foliar cover coated with ash increases exponentially (from ~10 to 90%) when grain size decreases from 500 to 90 µm, whether in dry or humid leaf conditions (Fig. 2). This relationship was established for a single ash mass load (~570 g m -2 ). For ash in the 267 intermediate size range, a higher load could result in enhanced retention of the particles, particularly along the primary and secondary leaf veins as these consist of less elastic tissues that can better absorb the kinetic energy of impinging ash particles of intermediate grain size.

270
However, for fine ash, we do not expect more retention to occur if tomato and chilli pepper leaves were exposed to higher loads because a large proportion of the uncovered foliage is comprised of leaves that, due to their steep angle, cannot retain ash particles efficiently. As 273 mentioned earlier, coarse ash particles tend to lodge primarily on leaf folds. Thus, their retention on foliage will likely be limited by the number of leaf folds. Overall, we anticipate that for ash load values >570 g m -2 , the exponential dependence of ash retention on ash grain 276 size will start to degrade and instead, a linear relationship would be a better model. The increased ash retention when grain size decreases is in accordance with the field observations of Miller (1967) after the 1963 eruption of Irazú volcano, Costa Rica, who found a higher 279 degree of retention of the smaller particles by crop foliage (alfalfa, maize, bean, beet, cabbage, carrot, pea, pepper, potato, radish and squash). Johnson and Lovaas (1969) and Witherspoon and Taylor (1970) reached a similar conclusion after dusting various crops 282 (alfalfa, maize, squash, soybean, sorghum, peanut and clover) with quartz powders differing in grain size (88-175 and 175-350, and 44-88 and 88-175 µm, respectively).
The fate of a solid particle falling from the atmosphere and hitting a leaf surface will depend 285 on how much of its initial kinetic energy is absorbed through tissue deformation (Vogel, 1989;Niklas, 1999;Benson, 2015). Ignoring aggregation processes and considering a constant particle bulk density, the coarser the particles, the larger their terminal fall velocity 288 and thus, kinetic energy (Dellino et al., 2005;Benson, 2015), simply reflecting that mass increases with grain size. If particles retain enough kinetic energy after impact, they can bounce back and be ejected off the leaf or deposited elsewhere (Gregory, 1961;Chamberlain, 291 1967; Starr, 1967;Chamberlain and Chadwick, 1972). Otherwise, they will settle on the upper side of leaves, although they may be subsequently displaced as new particles impinge the leaf experiment. Terminal fall velocity increases with grain size and is five times lower for 297 particles of 100 µm diameter (assimilated to the fine ash fraction) than for particles of 410 µm diameter (corresponding to coarse ash) (Table S4). This result suggests that the kinetic energy of the finest ash particles is ~10,000 times smaller than that of the coarsest material. The low 300 kinetic energy of fine particles probably explains why ash in the ≤ 90 µm size fraction produces a greater foliar cover compared to ash ≥ 500 µm (Fig. 2). In contrast, coarse ash particles with higher kinetic energy will tend to lodge on less elastic leaf structures, such as 303 primary and secondary veins and folds (Fig. 3). As mentioned above (section Material and methods), an inherent limitation of our experimental study is that the ash material did not contain the vesicular particles that are usually found in various proportions in ash fallout from 306 explosive eruptions. We speculate that the irregular shape of vesicular ash could enhance retention on foliage, perhaps even more so if the leaf surfaces are pubescent or wet. Thus, our measurements may be regarded as conservative estimates.

Influence of leaf pubescence on ash retention
On average, ash particles in the intermediate size range 125-500 µm cover ~25% more foliar cover in tomato than in chilli pepper (Fig. 2, Table S1). This is attributed primarily to the 312 presence of leaf hairs in tomato. Saebø et al. (2012) and Ram et al. (2012) demonstrated that dust accumulation on the foliage of various trees and shrubs is proportional to leaf hair density. Leaf hairs enhance dust collection area and capacity to absorb the falling particles' 315 kinetic energy. In addition, leaf pubescence may prevent particles from sliding off the leaf surface. By increasing friction on particles, leaf hairs counteract the gravity force generated by mass loading on the leaf surface which pulls a leaf downward (Smith and Staskawicz, 318 1977). In our experiments, ash ≤ 90 µm adhered to the tip of pubescent leaves with a steep inclination angle in tomato plants, whereas it barely encroached on the glabrous surface of chilli pepper leaves (Fig. 3). Previous field observations of ash-impacted crops also highlight 321 a stronger adherence of ash on pubescent leaves (such as barley, corn, tobacco, tomato and apple tree) and hairy fruits (such as peach, apricot, kiwi-fruit, strawberry and raspberry) (Miller, 1967;Cook et al., 1981;Sword-Daniels et al., 2011;Ligot et al., 324 2022). Witherspoon and Taylor (1970) concluded that the pubescent leaves of squash and soybean favour a uniform retention of quartz particles (88-175 µm). In contrast, the glabrous leaves of rose plants exposed to the 1963 eruption of Irazú volcano, Costa Rica, collected 327 little ash material (Miller, 1967).

Influence of humidity conditions at leaf surfaces on ash retention
Wetting of leaves prior to application of ash with an intermediate grain size of 90-500 µm 330 increased the foliar cover coated with ash of tomato and chilli pepper by 17 ± 5% and 31 ± 10%, respectively (Fig. 2, Table S2). We also noted that the ash deposit that formed on prewetted leaves appeared more homogeneous compared to that observed when the leaf surface 333 was dry (Fig. 3). Similarly, Miller (1967) reported during the 1963 eruption of Irazú that wet leaf surfaces facilitated retention of ash < 300 µm and formation of a homogeneous deposit.
Enhanced ash retention on wet leaves likely relates to the surface tension generated by water 336 molecules present on the leaf surface (Tabor, 1977;Israelachvili, 2011). Conversely, as plant leaves are hydrophobic (Bhushan and Jung, 2006), more water on leaves, such as after a heavy or prolonged light rain, could lead to formation of large water droplets able to erode 339 particle from the leaf surface, thereby reducing ash retention.

Modelling potential yield loss in tomato and chilli pepper plants exposed to ash
Our experimental results indicate that ~570 g m -2 fine ash can readily cover the upper side of 342 leaves (Fig. 2). Assuming an ash material comprised of spherical particles 90 µm of diameter and with a density of 2.54 g cm -3 (i.e. the density of phonolite), we calculated that a mass load as low as ~8.6 g m -2 can form a monolayer deposit on a leaf surface. While this estimate 345 represents an oversimplified situation, it is more than fifty times less the ash load (~570 g m -2 ) used in our experiment. Since fine particles are ubiquitous-albeit in various proportionsin ash fallout (Rust and Cashman, 2011;Costa et al., 2016), an ash coating on leaf surfaces is 348 likely to be the rule in vegetated areas affected by explosive eruptions. Importantly, the presence of solid particles on foliage exerts a shading effect, which reduces light interception (LI, dimensionless) by leaves (Thompson et al., 1984;Hirano et al., 1990). For example, 351 Hirano et al. (1991) measured a ~20% decrease in LI after treating mandarin tree leaves with only 4 g m -2 of road dust (0.1-100 µm). Similarly, deposition of 10 g m -2 of ash (0-100 µm) on cucumber plants led to a ~20% reduction in LI (Hirano et al., 1992).

354
Considering that LI drives net photosynthesis rate and thereby, total biomass production (Wilson, 1967;Biscoe et al., 1977;Monteith, 1977;Weraduwage et al., 2015), we contend that even a thin ash deposit on crop leaves can drive yield loss. Thus, the interference of ash 357 with LI provides an indirect mean to predict the potential crop production loss for ash mass loads below the threshold (~6-30 kg m -2 mass load) of direct mechanical damage to plants.
Although we did not measure LI in our experiment, this parameter can be inferred using the 360 following expression (Monteith, 1969): where k is the light interception coefficient (dimensionless). The temporal evolution of LAI

366
In light-limited situation, i.e. the other growth parameters (e.g. water and nutrient status) being optimum, the daily biomass accumulation by crop canopy (CBIOc, g m -2 day -1 ) depends on LI according to (Monteith, 1972;Hatfield, 2014): where Q is the incident radiation (MJ m -2 day -1 ) and RUE (g MJ -1 ) the radiation use efficiency. Representative values for Q in Belgium (10.6 MJ m -2 day -1 , warm temperate 372 humid climate, Solargis, 2022) and RUE are available from the scientific literature (Table S5). We consider two effects of ash on plant yield: reduction in LAI and premature biomass senescence. The former leads to lower accumulated biomass after formation of the ash 381 deposit, whereas the latter is responsible for a loss of biomass that accumulated prior to ash fall. We hypothesise that LAI reduction and biomass dying in crop plants exposed to ash is directly proportional to the percentage of foliar cover coated with ash deposits (Fig. 2), 384 presupposing that ash-affected leaves lose their ability to perform photosynthesis efficiently.
Based on this, and using Eqs. (1), (2) and (3) To illustrate our approach, we estimated % for tomato and chilli pepper plants exposed to 393 ~0.6 mm (~570 g m -2 ) of ash. We tested different ash size distributions and evaluated the influence of humidity conditions at leaf surfaces on ash retention. Two scenarios of plant exposure to ashfall were considered: one in which 25% of the plant growth period is 396 completed (i.e. 32 days after transplanting for tomato and 57 days after transplanting for chilli pepper), and one in which 75% is achieved (i.e. 97 days after transplanting for tomato and 172 days after transplanting for chilli pepper). The daily LAI evolution of tomato and chilli 399 pepper plants during growth was computed in R using published data (Fig. S6).
In our model, the entire plant canopy received the same amount of ash, although some leaves may be less exposed due to their position on the stem. As the ash mass load is low (570 g m -

402
2 ), we also considered that ash deposition on leaves neither halt plant growth nor production of new leaves (Neild et al., 1998;Ligot, 2022). On the day of the eruption, the LAI is reduced by an amount corresponding to the percentage of foliar cover coated with ash. On the following days, it re-increases as new leaves formation resumes at a rate similar to that before exposure to ash. If time permits, the LAI may reach a value identical to that of a plant that would not have received ash. The calculated temporal evolution of the LAI of tomato plant 408 that has completed 25% of its growth period when it receives ash (90-125 µm in diameter, mass load of ~570 g m -2 ) in dry conditions is illustrated in Fig. 5a. A similar temporal evolution of LAI is obtained for chilli pepper (Fig. S7).

411
The presence of ash on plant canopy may lead to premature leaf senescence (as reported by Miller, 1967;Neild et al., 1998;Ligot et al., 2022), impacting CBIOh (Eq. 3). To account for this effect, we subtracted the ash-coated leaf biomass from the total canopy 414 biomass, the latter being comprised of the leaves and stem. For tomato and chilli pepper plants, leaf biomass represents ~60% of canopy biomass (Kleinhenz et al., 2006;Elia and Conversa, 2012;Poorter et al., 2015). The leaf biomass fraction affected by ash can be 417 inferred from Fig. 1. Resolving Eqs. (1) and (2), the temporal evolution of CBIOc for tomato or chilli pepper subjected to ash can be predicted. Fig. 5b illustrates this for tomato plant exposed in dry conditions to ash deposition (90-125 µm in diameter; mass load of ~570 g m -2 ) 420 32 days after transplanting (i.e. at 25% of growth period). Since the leaf-to-canopy biomass ratio and percentage of leaf biomass covered with ash which dies are equal for both crops (Table S5, Kleinhenz et al., 2006;Elia and Conversa, 2012;Poorter et al., 2015), a similar trend is inferred for chilli pepper (Fig. 6). (CBIOc) (b) of tomato plant exposed to ~570 g m -2 of ash (size range: 90-125 µm) 32 days after transplanting (i.e. at 25% of the growth period) in dry leaf surface conditions. The hatched area represents the leaf biomass produced by the plant before the ashfall event and 429 which will undergo premature senescence after it. The ash covered leaf biomass is inferred from the leaf-to-canopy biomass ratio (i.e. 60%) and the percentage of leaf biomass covered with ash (i.e. 48% for tomato in dry leaf surface conditions, Table S1).

432
As detailed above, ash impact on CBIOh is modulated by different factors, including the LAI fraction that becomes photosynthetically inactive due to the presence of ash coatings on leaves (i), number of days elapsed between ash deposition and emergence of new leaves (ii), 435 leaf-to-canopy biomass ratio (iii), and percentage of leaf biomass covered with ash and which eventually dies (iv). Our model calculations revealed that crop growth period determines the relative importance of each of these factors in determining CYL%. For example, if 90 µm ash 438 affects tomato and chilli pepper plants in dry conditions at 25% of their growth period, CYL% is most sensitive to (i) and (ii), whereas for older plants that have completed 75% of their growth, (iii) and (iv) are the main factors driving CYL% (see Supplementary materials).

441
In order to assess the error on CYL% estimates, we applied a stochastic approach with 10,000 simulation runs using a random value for each of the four factors (as listed above) that can influence the final model output. We posited that the values taken by factors (iii) and (iv) 444 follow a gaussian distribution (Table S5), whereas variable (i) and (ii), which are always in the range 0-1 and positive, respectively, are described by a truncated gaussian distribution. Fig. 6 shows the uncertainties on CYL% as computed by fitting the first and third quartiles 447 around the median CYL% value for tomato and chilli pepper plants exposed to ash of different grain sizes, either in dry or wet leaf conditions. Calculations were repeated for plants that receive ash when at 25 and 75% of their growth period. For tomato, CYL% increases with 450 decreasing ash grain size (Fig. 6). Tomato plants at 25% of their growth may experience a 2-17% decrease in yield depending on grain size and humidity conditions at leaf surfaces. A significantly higher CYL% (0-42%) is anticipated when ash affects plants at 75% of their 453 growth. A similar pattern emerges for chilli pepper where CYL% varies between 1-17 and 0-46% when considering that the plant receives ash when at 25 and 75% of its growth period, respectively (Fig. 6). For intermediate ash grain sizes between 125 and 500 µm, the CYL% is 456 5, 3, 8 and 4% greater for tomato compared to chilli pepper when exposure to ash occurs at 25% of the growth in dry conditions, 25% of the growth in wet conditions, 75% of the growth in dry conditions and 75% of the growth in wet conditions, respectively. Towards using LAI as an impact metric for predicting potential yield loss in ash-affected 465 crops While deployment of field-based post-EIA will continue to enrich our understanding of ashloss of production relationships, progress is contingent on eruption occurrence, site 468 accessibility, limited field time, variations in environmental conditions and incomplete ranges of ash characteristics such as thickness and grain size (Jenkins et al., 2015). Here, we have shown, using established theories of plant-physiological processes (Monteith, 1969;Monteith, 471 1972), how empirical data from experimental testing can be transformed into quantitative insights for predicting potential yield loss in tomato and chilli pepper crops exposed to ash.
Changes in LAI and premature biomass loss in ash-affected crops are interpreted as dependent 474 on ash retention on leaves, a process influenced by grain size, plant traits and environmental conditions (Fig. 1). Here, we exclude the possible effect of ash surface composition on ash retention. As detailed in Eqs. (1), (2) and (3), crop yield depends on LAI and therefore, the 477 latter is regarded as an integrative impact metric. From this, we propose that LAI measurements in crop plants subjected to ashfall offer a new method for analysing crop vulnerability and assessing potential yield loss for ash mass loads below the threshold (~6-30 480 kg m -2 ) of direct mechanical damage to plants. The rapidly increasing ability to monitor crop characteristics, including type, LAI and biomass, using optical and radar earth observation data (Hosseini et al., 2015;Fang et al., 2019;Rosso et al., 2022) provides an unprecedented 483 opportunity to collect a spatially-and time-resolved information that can support the development of more realistic and more complete ash-loss of crop production relationships.
In order to unlock the full potential of LAI estimates for investigating the vulnerability of 486 crops to ash events, more knowledge on how ash coatings on leaves interfere with LI is required. In our model of potential yield loss in tomato and chilli pepper (Fig. 6), we equated LAI reduction with the foliar cover percentage covered with ash. In essence, this means that 489 an ash deposit on leaves renders light interception inoperative. This may not always be the case because LI by a crop canopy is determined not only by the LAI of the species, but also by the light absorption characteristics of the leaves (Liang et al., 2012), here modified by the ash 492 deposit. Further laboratory investigations can generate the empirical observations needed to better constrain the changes in LI in relation to the characteristics (thickness/mass load, grain size, albedo) of the ash material deposited onto the leaf surface.
The evolution of LAI following an ash deposition event (Fig. 5a) was modelled by assuming that ash-affected plants will grow new leaves after a set period of time. Our analysis showed that CYL% is sensitive to this parameter, therefore requiring adjustment depending on crop 498 type (Klepper et al., 1982). We also note that many crops (including major ones such as wheat; Hay and Porter, 2006)  if upper leaves partly shield the surfaces of leaves located below them from direct exposure to ash. Thus, the effect of ashfall on crop LAI hinges both on plant growth characteristics and timing of the volcanic eruption. We considered in our model that an ash deposit induces 507 premature leaf senescence, in agreement with field observations (Miller, 1967;Neild et al., 1998;Ligot et al., 2022). While this process probably relates to leaf chlorosis due to LI reduction (Bilderback 1897; Mack, 1981;Ligot et al., 2022), its 510 temporality and precise mechanism remain unclear. New experimental investigations with various crop plants will help to better constrain the proportion of leaf biomass affected by ash which will be subjected to premature senescence. 513 We have highlighted that grain size, leaf pubescence and humidity conditions at leaf surfaces control ash retention, which in turn drives LAI reduction. Other factors may influence ash retention. For example, leaf microstructural features such as stomatal density and presence of 516 a waxy epicuticle have been shown to influence retention of non-volcanic dust particles (Saebø et al., 2012;Zhang et al., 2017). In addition, in the natural environment, wind-and rain-driven erosion processes can remove ash deposited on foliage. Conversely, light rain may 519 induce crusting of ash, prolonging its residence time on leaves (Miller, 1966;Ayris and Delmelle, 2012;Le Pennec et al., 2012;Ligot et al., 2022). The significance of these environmental variables in controlling ash retention time by leaves has never been assessed 522 quantitatively, calling for further field and experimental investigations linking ash residence time on plants and impacts.
Finally, our approach for modelling production loss in tomato and chilli pepper exposed to 525 ash neglects impact to flowers or harvested plant parts, and assumes that light interception is the main variable governing plant growth. While this is true in our study where water and nutrient supply were never limited, more stringent conditions may be encountered in crop 528 fields subjected to ashfall. For example, an ash layer on the ground may alter water and gas movements into and through the soil and surface runoff (Ayris and Delmelle, 2012; Neslon, 2013;Tarasenko, 2018), in turn impacting the soil water balance. A better comprehension of 531 the side effects of ash deposition on the soil plant-system is needed in order to identify the primary mechanisms driving the short-and long-term consequences for crop production.

534
Our study highlights the usefulness of conducting experimental measurements to supplement observations obtained from post-EIA. It provides a new perspective into the volcanic and nonvolcanic factors that control ash impact on crops. The experimental results obtained for 537 tomato and chilli pepper plants demonstrate that ash retention on leaf surfaces increases with decreasing grain size and is enhanced when leaves are pubescent and wet. We also showed that, for a given ash mass load (~570 g m -2 ), the leaf surface percentage covered with ash is an 540 exponential decay function of grain size of which the parameters are influenced by leaf pubescence and humidity conditions at leaf surfaces. Thus, we conclude that the proportion of fine material in ash fallout is an important hazard metric for assessing risk to crops. The corollary to this finding is that relying on ash thickness (or mass load) alone to anticipate crop damage from ash is inaccurate and possibly misleading.
Using the empirical relationship linking ash retention to ash grain size and equating ash 546 retention with LAI reduction, we have developed a novel model framework to predict CYL%.
This approach identifies LAI as a promising impact metric that can be quantified for assessing crop production following an ashfall event. LAI is commonly retrieved via remote sensing 549 measurements. The rapid deployment of new satellites allows data collection at increasingly high spatial and temporal resolution (for example, the European Space Agency's Sentinel-2 mission), paving the way for estimating LAI at the crop field scale. Additionally, the 552 technology gives access to FPAR, i.e. the fraction of the solar radiation absorbed by live leaves for the photosynthesis activity, which should also record a reduction in light interception for leaves covered with ash. We anticipate that tapping into satellite-derived 555 measurements will considerably improve our quantitative understanding of crop vulnerability to ash fallout. However, for exploiting their full potential, field-and laboratory-based validations are required, including experiments aimed at constraining LI/LAI reduction in 558 relation to ash retention and characteristics. Acquiring this knowledge will significantly enhance our capacity to estimate ash-related risks to crops accurately. Governments and payout agencies need such assessments in order to develop and implement effective risk 561 reduction strategies for ashfall damage to crops in volcanically active agricultural regions.

Code availability
The Image J macro to analyse the plant photos and estimate the foliar cover coated with ash 564 and the R script to compute the daily tomato and chilli pepper LAI, LI, CBIOc and CYL% are available on GitHub (https://github.com/NoaLigot/ImageJ-macro.git and Scientifique