Survival and efficacy of entomopathogenic nematodes on exposed surfaces

Entomopathogenic nematodes (EPN) species differ in their capability to withstand rapid desiccation (RD). Infective juveniles of Steinernema carpocapsae are a better adaptable and tolerant than Steinernema feltiae or Heterorhabditis bacteriophora as, an optimal RH of > 90% is required by S. feltiae and H. bacteriophora while maintaining RH equivalent to 74% could sustain survival of S. carpocapsae under RD. Our findings from infectivity suggest that following application, shrunk IJs are acquired passively by the larvae, probably rehydrate and resume infection within the insect gut. Water loss rate is a key factor affecting survival of S. carpocapsae on exposed surfaces. The present study provides the foundation for characterizing mechanism of rapid rate of water loss in EPN. ATR-FTIR is a rapid and reliable method for analysis of water loss. Changes in peak intensity was observed at 3100–3600 cm−1 (OH bonds of water), 2854 cm−1 (CH stretching of symmetric CH2, acyl chains), 2924 cm−1 (CH stretching of anti-symmetric CH2, lipid packing heterogeneity), 1634 cm−1 (amide I bonds) indicate major regions for hydration dependent changes in all EPN species. FTIR data also indicates that, S. carpocapsae contains strong water interacting regions in their biochemical profile, which could be an influencing factor in their water holding capacity under RD. ATR-FTIR were correlated to water content determined gravimetrically by using Partial Least square –Regression and FTIR multivariate method, which could be used to screen a formulation’s potential to maintain or delay the rate of water loss in a rapid and efficient manner.


Results
Survival and rate of water loss indicates S. carpocapsae is better tolerant to RD in comparison to S. feltiae and H. bacteriophora. EPN species differ in their response to RD at 85% RH ( Fig. 1-A1).
More than 80% of S. carpocapsae survived in comparison to complete mortality observed in S. feltiae and H. bacteriophora at 2 h into RD. Hence, rapidly desiccated S. carpocapsae survival was significantly different from S. feltiae and H. bacteriophora (p < 0.0001) at 2 and 4 h. However, S. feltiae and H. bacteriophora were not significantly different from each other at 0, 2 and 4 h (p = 1.0, p = 0.99 and p = 1.0) respectively.
Water content of S. carpocapsae, S. feltiae and H. bacteriophora varied with time and among species. The water content of EPN species determined by gravimetric measurements was significantly different and varied with time ( Fig. 1-A2 and p < 0.0001). All EPN species lost ca. 50% of their water content within the first hour of exposure to RD. After 1 h, S. carpocapsae had the lowest reduction in water content (54%), H. bacteriophora had  (HB)) rapidly desiccated at 85%RH. Survival (%) (A1) and gravimetric rate of water loss (A2) at 85%RH. (A3-A5) Comparison of FTIR spectra depicting the water peak of rapidly desiccated S. feltiae, S. carpocapsae, H. bacteriophora at 85%RH for 0, 2 and 4 h. Inlet in each figure represents the peak of OH (water)/NH bonds. The numbers inside the spectra represents the bands that were strongly affected due to different humidities (see Table S1 for assignment of functional groups). www.nature.com/scientificreports/ the highest reduction (69%). Additional exposure for 1 h at 85%RH resulted in further reduction of 30-33% and 25-27% water content in H. bacteriophora and S. feltiae respectively, whereas S. carpocapsae lost only 10% of water content. In addition, at 4 h, S. carpocapsae had a fourfold higher water retention (approx. 30% water content) and significantly different in comparison to S. feltiae (ca.7% water content) and H. bacteriophora (p < 0.001). Among the tested EPN species, S. carpocapsae was able to adapt and maintain water content reduction from 1 to 4 h of RD at 85% RH as also indicated by its survival.
Survival of S. carpocapsae under low humidity. In vitro survival of rapidly desiccated S. carpocapsae at lower humidities of 85, 74, 64 and 43% RH was estimated. The experiment was repeated four times with five technical replicates and the humidity was maintained as mentioned in Table 1. Response of S. carpocapsae survival varied with different humidities and was significantly different from control (p < 0.05) ( Fig. 2-A1). Survival of S. carpocapsae IJs varied with time and was found to be significantly different (p < 0.001). At 85% and 74% RH, S. carpocapsae exhibited the highest survival (> 80% and > 75% respectively) while at 43% humidity the sharpest decline in survival to 20% within 4 h was observed. This indicates lower tolerance capability of S. carpocapsae towards harsh humidities below 60% RH. Two-fold reduction (20%) in survival of S. carpocapsae was observed at 43% RH in comparison to 55%RH with an observed 58% survival (Fig. S1). The least tolerable humidity range for S. carpocapsae at 23 °C was found to be between range of 55-43% RH with a significant difference within this humidity range (p < 0.0001). Gravimetric water content of S. carpocapsae at different RH were significantly different with time (p < 0.001) ( Fig. 2-A2).Water content in S. carpocapsae reduced to approximately 40, 25, 18 and 1% within 2 h of exposure at 85, 74, 64 and 43 RH respectively. At 85%RH, water content in S. carpocapsae reduced gradually from 40 to 12% and at 74% RH, reduced from 25 to 8% over the time of 2-6 h. In comparison, water content at 64 and 43% RH reduced drastically to 20% and 1% within 2 h of exposure.
Efficacy of S. carpocapsae at different humidities. Nematode exposed to RD, followed by introduction of host and its influence on efficacy of IJs was assessed. After application of IJs on leaf and exposure to 85%, 64% RH, larvae starved for 1 h were added at 0, 2, 4, 6, and 8 h (Fig. 3-B1-B3) and sealed with mesh screen plates.
Relative humidity had an impact on the efficacy of S. carpocapsae. The efficacy against late fourth larval instars of S. littoralis decreased with decrease in RH and observed decreased efficacy with increase with time. Although no statistical significance in mortality at 85%RH was observed ( Fig. 3-A1), it is critical to note the nematode survival at the time of larval addition and correlate it to the mortality observed ( Fig. 3-A1 and A3). A significant (p < 0.0001) impact in efficacy was observed at the lower humidity ( Fig. 3-A2). At optimal humidity, complete mortality was observed in comparison to 72 and 64% insect mortality observed at 85 and 64%RH for 0 h. However, with time, mortality reduced drastically from (ca.) 64 to 20% between 2 and 4 h at 64%RH, while decline in survival from remained consistent ( Fig. 3-A2 and A3). In addition, no mortality was observed at 6 and 8 h of desiccation, while survival remained at ca. 40%.

ATR-FTIR as a rapid, efficient alternative tool for water content determination. EPN species
were prepared for ATR-FTIR according to San -Blas et al. 37 . FTIR fingerprint spectra of EPN species varied in response to RD at 85% RH ( Fig. 1-A3-A5).
The peak at 3100-3600 cm −1 generally represents water and NH bonds in nematodes. The reduction of peak height and intensity at this region in EPN species represents the effect of hydration of nematodes. In accordance with the gravimetric water loss, S. carpocapsae accounted for the slowest reduction while H. bacteriophora had the highest reduction in water content within 4 h. In addition, in FTIR, several regions such as 2854 cm −1 (acyl chains), 2924 cm −1 (lipid packing), and 1634 cm −1 (amide bonds) indicate major regions for hydration dependent changes in RD was evident from two-way ANOVA test (Table S1). Interestingly, spectral regions assigned to lipid packing changes (Region 4 in  FTIR univariate analysis. FTIR spectra at peak 3300 cm −1 (indicative for NH bonds and region of 3100-3600-indicative for water) record a continuous decrease in peak height ( Fig. 4-A1) while the dry spectra of RD  Table S1 for assignment of functional groups). Survival (%) (n = 360) and gravimetric water loss (%) (n = 156) were arcsine square-root transformed and subjected to comparison of standard least squares by REML. When ANOVA F was significant (p < 0.05), means were compared. Mean was compared using Tukey test. Different alphabets represent significant difference. www.nature.com/scientificreports/ nematodes at 3100-3600 cm −1 indicate that peak differences observed at differing humidities were an effect of water ( Fig. 2-A3, indicated by black line). Hence, FTIR data was interpreted by Univariate analysis. This univariate approach indicated that, at higher humidity (85 and 74%RH) the relative water content was two-one fold higher in comparison to gravimetric measurements respectively ( Fig. 4-A1 and A2). To confirm, we identified if the fit of the data for high humidity was acceptable ( Fig. 4-A3). The fit of the data (R 2 ) ranged between 0.78-0.55 from 3560-3100 cm −1 while fit of data at 3301 cm −1 was 0.58. In addition, fit of data (R 2 ) obtained from area under curve (AUC) compared to gravimetric measurements was 0.60. These results indicate that the effect of water loss cannot be completely explained by the region of 3600-3100 alone and changes in other spectral regions in IR (i.e.) 4000-400 cm −1 (Fig. 2-A3-A5) should be considered for predicting water content as well.
FTIR multivariate analysis. Exploiting the data obtained from whole FTIR spectra (4000-400 cm −1 , Fig. 2-A3-A6) and model water content with corresponding gravimetric measurements as reference to calculate multiple partial least squares-regression models to perform their validation.
The spectral regions and quality indicator (R 2 -coefficient of determination) for cross validation was used to calculate the PLS-R model. The average number of PLs components were inferior to 7 and the coefficient of determination was greater than 0.93, confirming the quality of the selected models. Hence, the data arising from the predictions were plotted against the gravimetric water content and FTIR univariate (Fig. 5) and the comparison of actual vs predicted values obtained from Gravimetric estimations and PLS-R predicted values respectively are indicated in (Fig. S5). It is important to notice that, the predictions arising from FTIR PLS-R model fits very well with the experimental gravimetric measurements (R 2 > 0.93) highlighting the requirement of FTIR multivariate.  www.nature.com/scientificreports/

Discussion
We evaluated and correlated water loss through RD and survival of three EPN species. Two methods were applied: gravimetric, and spectral-based ATR-FTIR. Gravimetry is the standard reference method for water-content analyses, whereas ATR-FTIR enables rapid measurements of biological specimens 32 , and is the best approach to studying hydrated and dried samples, such as cells 33 . FTIR has been extensively used to study hydrationdependent changes in desiccated biological materials, to identify, e.g., the survivability of pollen 34 , the role of trehalose in preconditioned Caenorhabditis elegans 35 and the ability of lipid bilayers to increase their affinity for trehalose during slow desiccation 36 . However, in EPNs, FTIR application has been limited to EPN species identification and analyses of biochemical composition 37,38 . In this study, identifying the rate of water loss in nematodes was a basic prerequisite to evaluate the effect of RD on nematode physiology. Hence, ATR-FTIR was used to identify the water loss from nematodes. Moreover, we wanted to correlate the gravimetric and FTIR measurements based on the biochemical profile of EPN species during RD at different RH and times, and on differences in the 3100-3600 cm −1 peak assigned to hydroxyl (-OH) vibrations of water 39,40 .
The tolerance of different EPN species to RD conditions was compared, and Heterorhabditis and S. feltiae were found to be poor survivors. These findings are similar to those of Liu et al. 11 , where their survival was estimated to be 15 min at RH ≤ 88%. Other studies have shown that the EPN species studied in this work differ in their survival rates under RD 14,29,30,41 . The better survivability of S. carpocapsae found here under RD at moderate RH of 74-85% has been noted previously as well 14,26,29,42,43 . Under the environmental conditions of RD used in this study, > 60% water loss within the first 2 h was detrimental to the survival of both S. feltiae and H. bacteriophora, in accordance with Patel et al. 41 . In contrast, S. carpocapsae was able to survive a water loss of > 50% in the first 2 h. FTIR spectra also indicated a decrease in absorbance of the water peak at 3300 cm −1 at different time points in accordance with RH. PCA of the FTIR data for S. carpocapsae, S. feltiae and H. bacteriophora also indicated their differential distribution with time under RD at 85% RH. S. feltiae and H. bacteriophora clustered together, while S. carpocapsae was grouped separately, indicating a different response to RD.
The advantageous survival capacity of S. carpocapsae has been attributed to factors such as water permeability of the cuticle and other physiological adaptations, among them epicuticle thickness and chemical composition 42,[44][45][46] . Interestingly, S. carpocapsae had thinner cuticle and epicuticular layers (273 nm and 18 nm, respectively) than S. feltiae (374 nm and 33 nm, respectively) 41 ; data for Heterorhabditis were not found in the literature. Somvanshi et al. 47 and Yaari et al. 17 showed that EPN species with higher desiccation tolerance, such as S. carpocapsae, are characterized by the constitutive expression of genes known to be involved in desiccationtolerance mechanisms. Recent findings from Gade et al., also indicate that, related organisms such as C. elegans possess a common survival framework in response to varied abiotic stress 48 . These results relate to preconditioning while our results are based on rapid desiccation. Thus, the results in the present study imply an intrinsic www.nature.com/scientificreports/ mechanism controlling the rate of water loss in S. carpocapsae. Hence, the higher tolerance of S. carpocapsae IJs to RD could also be attributed to bioprotectant processes. However, this hypothesis requires further investigation on the molecular and biochemical changes occurring during RD. Our results showed that the ability to manage water-loss rates depends strongly on the surrounding RH conditions. The IJs of S. carpocapsae were able to gradually reduce their rate of water loss at higher humidities (85% and 74% RH) from 0 to 6 h. However, at lower humidities (64% and 43% RH), poor survival was observed within 4 h, indicating that these IJs could not retain their internal water under these low RH conditions. This implies the requirement of a minimum range of time for controlling the rate of water loss and maximizing the intrinsic control and protective mechanisms. Findings of survival and measurements of the rate of water loss by both gravimetric and FTIR measurements in this study indicated that the first 4 h of RD exposure are critical for protective mechanisms. Differences in water content of S. carpocapsae, S. feltiae was observed by Patel et al. 41 from the current study, within 1 h and 30 min of RD at 85% RH respectively. The variation in results could be due to: (a) different methods of water-content estimation; (b) the substrate's ability to hold moisture. Thus, substrate also plays a vital role in influencing the survival of EPNs under RD.
Our FTIR results indicated that S. carpocapsae can maintain water at 85% RH, in comparison to S. feltiae and H. bacteriophora thus reiterating the water-holding capacity of S. carpocapsae at high humidity (85-74% RH). An interesting observation from FTIR of EPN species was in the region of 2854, 2924 cm −1 . As suggested by Erkut et al., (2011) this region is attributed to lipid chain packing heterogeneity (2924 cm −1 ) and reduced lipid-packing density (2854 cm −1 ). These regions in S. carpocapsae were unaffected by water loss at 85% RH, even though gravimetric water content was ca 38%. This suggests strong water-interacting regions in the biochemical profile of S. carpocapsae, such as the chemical composition of lipid-rich epicuticle layer and the glycoproteinrich surface coat, which could all be influential factors in the high tolerance to RD [49][50][51] . Hence, if these regions are unaffected, it is indicative of the effect of water permeability. Selvan et al. 52 identified differences of chemical composition (i.e.) higher saturated lipids in S. carpocapsae in comparison to higher unsaturated lipids observed in S. feltiae. This could contribute to water retention in S. carpocapsae as membranes made of saturated fatty acids are less permeable to water and salts. However, this hypothesis requires further research. Thus, FTIR in particular enables to compare chemical components/functional groups of EPN under the same state of desiccation. Celino et al. 53 identified and correlated gravimetric and FTIR data of plant fibers using univariate and multivariate methods. In the univariate method, relative water content was estimated solely as a representative of variations arising from the absorbance of hydroxyl groups at 3100-3600 cm −146 , i.e., spectral data obtained from one peak that is characteristic of vibration of hydroxyl bonds in water (at 3300 cm −1 ). Hence, the large difference in peak height observed for dry and RD S. carpocapsae could indicate interactive regions of hydrogen bonds. Hence, we measured the peak height of the samples after subtraction from the dry spectra. This enabled identifying changes induced only by loss of hydroxyl groups. The comparison of peak height and gravimetric measurements yielded differences that are best explained by Park's model, attributing them to the ability of water molecules to form clusters at higher humidity 54 . Thus, the whole IR spectrum of RD nematodes was also impacted by changes in humidity and time, from spectral regions 1600-500 cm −1 , indicating that a multivariate approach is suitable for accurate monitoring of water content, as also noted by Celino et al. 53 . Our analysis inferred that the multivariate method is better correlated to gravimetric measurements than the univariate method. This could be attributed to the consideration of all regions from 4000 to 400 cm −1 to obtain better predictive models to quantify water content. In addition, the R 2 value of 0.93 in the cross-validation of the obtained data by PLS-Regression method indicated a very high correlation between predicted and gravimetric measurements.
Another interesting observation is that insect mortality due to nematode infection accrued, even after the first 4 h at 85 and 64% RH on the leaf surface, whereas IJs tended to cease movement and shrink within 30-40 min of exposure to low RH. Technically, even one IJ could cause mortality however; we observed a drastic reduction in mortality at 64% that might indicate that low humidity condition has an effect on the nematode efficacy irrespective of the concentration. We therefore presume that the positive infectivity is due to rehydration of IJs in the insect gut and resumption of infectivity. These findings are in agreement with Glazer 14 , who identified nematode shrinkage after 45 min on the leaf surface while retaining efficacy, indicating route of entry through consumption of treated leaves by the insect. This ability of desiccated IJs to rehydrate inside the insect gut and resume infectivity was also reported by Wojicik and Georgis 55 . Recent research on Diptera has also indicated the ability of IJs to infect larvae by passage to the haemolymph from the gut through the peritrophic membrane 56 . This would indicate that motility of rehydrated IJs inside the insect gut is critical for successful infection. Hence, at 64% RH, no insect mortality was recorded despite the 40% survival rate of IJs; we assume that these IJs were not able to enter the haemolymph or were rendered inactive during rehydration.
The results of the current study lead to the interesting assumption that host behaviour on the foliage is crucial to appropriately pair the right method of EPN application, given that infectivity is still positive under non-active penetration (by ingestion) of EPNs. In turn, this could mean that requirements for formulations have to be preestablished, depending on the host. For example, in the case of active, polyphagous feeders such as Spodoptera, a formulation requirement would be to maintain a steady rate of water loss with rapid rehydration. This is in contrast to other types of foliage pests, such as leaf miners or sucking pests, where it is necessary to maintain the activity of IJs to ensure host tracking and successful infection.

Conclusion
Overcoming the low viability and erratic infectivity of EPNs on foliage through formulations has been a focal point of research since 2000 24,[57][58][59][60][61][62][63] . Here, we focused on the physical parameters of RD as well as its effect on IJ efficacy on exposed surfaces. ATR-FTIR is a rapid and reliable method for analysis of water loss. Changes in peak intensity were observed at 3100-3600 cm −1 (OH), 2854 cm −1 (symmetric CH stretching CH 2 , acyl chains), www.nature.com/scientificreports/ 2924 cm −1 (antisymmetric CH stretching CH 2 , lipid packing) and 1634 cm −1 (amide I bonds), indicating major regions for reflecting hydration-dependent changes in all EPN species. ATR-FTIR measurements were correlated to water content determined gravimetrically using partial least square regression and FTIR multivariate method, and this could be used to screen a formulation's potential to maintain or delay the rate of water loss in a rapid and efficient manner. The present study provides the foundation for characterizing rapid rates of water loss. Our observations indicate that following exposure to RD, IJs cease moving and shrink within 20-40 min; however, insect mortality due to nematode infection continues to occur, even after 2 h of exposure. Our findings suggest that shrunken IJs are acquired passively and probably rehydrate and resume infection within the insect gut. Hence, it can be extrapolated that although active infection processes (of IJ) are a sought-after trait in EPNs, on foliage, host behaviour also influences the dynamics of the infection process. From this perspective, the nature of EPN behaviour during RD is crucial for infectivity. Our findings provide the basis for the development of new formulations that will enhance nematode survival and activity during the critical period required to ensure efficacy. Further fundamental biochemical and molecular research is required to maximize formulation functionality for survival and infectivity of EPNs on foliage.

Materials and methods
Nematode source. EPN response to RD was characterized by studying the commercially available lines of Steinernema carpocapsae, Steinernema feltiae and Heterorhabditis bacteriophora obtained from e-nema (e-nema, GMBH, Schwentinental, Germany). IJs were reared in last instar larvae of Galleria mellonella according to Stock and Kaya 64 . The cadavers were transferred to modified White trap 65 and incubated at 23 °C. The emerging Infective Juveniles (IJ) were collected and stored at 8 °C for Steinernema and 14 °C for Heterorhabditis until used further for bioassays. Every 2 weeks fresh stock of IJs were propagated.
In-vitro EPN survival at high relative humidity. Survival of EPN species (S. carpocapsae, S. feltiae and H. bacteriophora), rapidly desiccated at 85% humidity was estimated and compared. Approximately 5000 IJs in 1 ml of each of the 3 nematodes lines were vacuum filtered on filter paper (Whatman no.1, 55 mm Diameter), dried under laminar airflow (LAF) chamber for 10 min, to remove excess water. Samples were transferred to desiccator at 85% RH (maintained with saturated salt solutions as outlined in Table 1 66 ), 23 °C and recorded continuously for humidity by data logger (LOG32TH and SSN23E). Nematodes suspended in saline and incubated at 23 °C served as control. Samples were collected at 0 (immediately after filtration), 2 and 4 h of RD. Samples from filter paper (approximately 0.6 cm × 2 cm) were cut, suspended in saline solution and incubated overnight for assessment of nematodes survival rate. This was done by random counting the number of live or dead among approximately 200 IJs under binocular (Olympus, SZH10, 30× magnification). IJ were scored alive on active movement and on response to probing. Data was used to calculate survival rate as percentage of total number of nematodes. The experiment was repeated twice with 5 replicates per treatment.

Survival of S. carpocapsae under low humidity.
In-vitro survival of rapidly desiccated S. carpocapsae at lower humidities of 85%, 74%, 64% and 43% RH was estimated. The nematodes samples were prepared as outlined in previous section. Samples from filter paper (0.6 cm × 2 cm) were cut, suspended in saline solution and incubated overnight for survival rate. Survival rate was determined as outlined previously. The experiment was repeated four times with five replicates per treatment.
Evaluating EPNs water content by gravimetric method. Water content of EPN species during RD was estimated using Gravimetric method as described by Erkut et al. 35 with slight modifications. IJs at an amount of 50,000 IJ suspended in 120-150 µl saline were vacuum filtered on 1.3 cm diam. filter paper and dried under laminar airflow for 20 min. The obtained samples following this procedure were regarded as 0 h. samples were weighed (0 h) and transferred to desiccator maintained at specified humidities (85, 74, 64, 43% RH). Control treatments consisted of 120-150 µl of saline solution that were processed as mentioned above to remove the effect of residual water in filter paper. vacuum filtered and dried. All samples and controls were weighed at various time points (1, 2, 3 and 4 h at 85% and for S. carpocapsae at lower humidities-2, 4, 6, 24 h at 85, 74, 64, 43% RH). Dry weight was estimated by oven drying at 90 °C for 1 h. Weight of water were calculated by subtracting the dry weight of nematode samples from the estimated weight at different time points. Further, the weight of the control samples were also negated to remove the effect of residual saline on the water loss. Hence, the water content of nematodes estimated in weight (mg) and was expressed in percentage (%) as Normalized body water content (NBWC) using the following formulae: where, "W t " represents weight of samples at different time point (e.g.: 0, 2, or 4 h). "W 0 " is the initial weight at 0 h. The experiment was repeated 3 times with three technical replicates. www.nature.com/scientificreports/ ATR-FTIR (Bruker, Germany), pressure clamp was applied. The spectra were collected from 4000 to 400 cm −1 , 4.0 cm −1 spectral resolution and 32 scan. Spectra was also collected for oven dried samples at 90 °C for 1 h. The experiment was repeated twice with three technical replicates per sample.

Attenuated total reflectance-Fourier transform infrared spectroscopy (ATR-FTIR
FTIR spectra pre-processing. All the spectra were recorded to the background spectra. For the multivariate analysis, the raw spectra were automatically corrected for ambient CO 2 , smoothed by adjusting the data spacing at 16 cm −1 . All these treatments were achieved using integrated functions of the instrument (Nicolet IS5, Bruker Optics, Germany). Pre-processed spectra were exported as text files for file format modification on Microsoft Excel. The spectral data were further analysed by performing multivariate analysis Principal Component Analysis (PCA) using the JMP15 53 .
FTIR univariate analysis. The spectra of RD nematodes at the region of 3600-3100 cm −1 were subtracted from the dry nematodes. The resultant peak at 3300 cm −1 was measured for peak height using the peak height tool. This region is considered to be representative of OH bonds. All these changes were carried out in the Omnic software. The measured peak heights were plotted against the gravimetric water content. The peak area of region 3100-3600 cm −1 was measured by origin software (OriginLab Corporation, Northampton, MA, USA) version 8 by fit single peak with built in guass function after baseline correction. The baseline was corrected by fitting method under exponential curve and by finding anchor points for detecting the baseline.
FTIR multivariate analysis. Calibration and cross validation was conducted using Partial least squares-regression (PLS-R). Firstly, the pre-processed infrared spectra were Partial least squared against the reference values obtained from Gravimetric water content estimation. At this step, a formula is generated for each wavelength to equate to the gravimetric measurements obtained. This step also allowed building models for prediction of formulae based on the number of correlated factors by NIPALS and with number of factors ranging from 5 to 7. The number of factors identified were based on percentage of variations that was able to explain the differences between gravimetric measurements (Y response) and FTIR wavelengths (X response). The number of factors identified was 5 based on the percent variations. This factor was able to explain 94% variations in FTIR based on 95% variation in gravimetric measurements. Secondly, a cross validation was performed by regression method using the lasso estimation and holdback of 0.33% of the samples (i.e.) 11 spectra were randomly selected by the software to perform the cross validation. This procedure allowed finding models with high correlation (R 2 ).

Efficacy of S. carpocapsae at different humidities.
Nematodes were assayed for their activity against the 4th instar of the foliage feeder Spodoptera littoralis (Egyptian cotton leafworm) larvae. Insect colony was continuously maintained as described in Birnbaum et al. 67 . The larvae were used 16-19 days post hatching to ensure homogenous development. The experiment was carried out on 55 mm petri plates with holes to insert the petiole of the cotton leaf. To keep the leaves fresh, the tip of the petiole was wrapped with moist cotton (Fig. 3-B1-B3). Each leaf was applied with 5000 nematodes spread with loop, allowed to dry under hood for 15-20 min. Control plates were sealed with regular petri lids, wrapped with Parafilm and sealed. Mortality was assessed after 48 h. and gently transferred to desiccators maintained at 75% RH and 55% RH at 24 ± 1 °C. A shift in + 10%RH was recorded due to the presence of leaf. The nematodes were exposed to each RH for 0, 2, 4, 6 and 8 h before exposing it to the insect larvae. In addition, nematodes were also observed for their behaviour of shrinking at the time of larval addition ( Fig. 3-C1 and C2). Five instar larvae were added, closed with mesh screened plates and sealed with Parafilm. The plates were maintained in the desiccators and incubated at 24 ± 1 °C. Insect mortality was determined after 48 h. Each treatment contained five technical replicates (dishes). An extra replicate was monitored for evaluating the survival of the nematodes. These samples were evaluated for nematodes' survival following overnight incubation. Experimental protocols involving plant materials was conducted in accordance with institutional, national, and international guidelines and legislation. The Akala variety cotton plants were grown under greenhouse conditions for 3 months and used for experiments and permissions were obtained from the Agricultural Research Organization. Statistical analysis. All statistical analyses were undertaken using JMP version 15 pro (SAS Institute Inc.).
Comparisons of survival and gravimetric water content bioassays were carried out on the arcsine-transformed proportions of live nematodes and normalized body water content (%). They were subjected to an analysis of standard least squares by restricted maximum likelihood (REML) followed by Tukey's HSD test for multiple comparisons among means. Survival (%) (n = 90) and gravimetric water loss (%) (n = 72) were arcsine squareroot transformed and subjected to two way ANOVA and comparison using standard least squares by REML method respectively. When ANOVA F was significant (p < 0.05), means were compared. Mean was compared using Tukey HSD test. Different alphabets represent significant difference. Insect mortality data in percentage, was corrected for mortality in the control using Schneider-Orelli's formula (Püntener, 1981), arcsine transformed and subjected to one-way ANOVA. For nematode survival on leaf, the survival values were arcsine -transformed and subjected to standard least squares by restricted maximum likelihood (REML) followed by Tukey HSD test. Nematode survival (%) (n = 100), insect Mortality (%) of 85%RH (n = 50) and 64%RH (n = 50) were arcsine square-root transformed, subjected to two-way ANOVA and one-way ANOVA respectively. When ANOVA F was significant (p < 0.05), means were compared. Mean was compared using Tukey HSD test. Different alphabets represent significant difference. To identify the wavelength that were strongly affected by humidity, FTIR spectral wavelength data at different humidities and time points were subjected to two-way ANOVA with FDR correction applied to p values. www.nature.com/scientificreports/ Ethics approval. All experimental protocols involving plant materials was conducted in accordance with institutional, national, and international guidelines and legislation under the methods section of the main manuscript file.