Due to climate change, the constant rise of ambient temperature is one of the major global issues and has devastating impacts on crop growth and productivity, consequently, in food security. Results from various climate model simulations predict that the global average temperature could be between 1.1 to 5.4°C warmer in 2100 than it is today (IPCC, 2019). Heat stress causes a manifold adverse impact on the growth, development, physiological processes of plants (Wahid et al., 2007; Hasanuzzaman et al., 2013; Hassan et al., 2021; Jagadish et al., 2021). The plant employs varying levels of adaptation, avoidance, acclimation, and tolerance mechanisms to cope with heat stress via physical, physiological, biochemical, and molecular strategies, including the use of ion transporters, proteins, osmolytes, antioxidants, and other factors involved in signaling and transcriptional regulation (Hasanuzzaman et al., 2013; Hassan et al., 2021; Jagadish et al., 2021; Zhao et al., 2021). Plant responses to abiotic stress have been studied widely for the last few years, and the role of microbes on plant stress responses has also been given attention in recent years. In particular, plant species commonly associated with microbial symbionts, known as plant microbiome, may influence responses of the host plant to environmental stimuli, including heat stress.

Microorganisms are ubiquitous, and all plants are colonized by various types of microbes which play important roles in plant ecology and physiology (Rodriguez et al., 2009; Turner et al., 2013; Compant et al., 2019; Dastogeer et al., 2020a). For example, AMF (arbuscular mycorrhizal fungi) present in the roots of 80% of terrestrial plants (Brachmann and Parniske, 2006) and microbial endophytes (both fungal and bacterial) are highly diverse and live asymptomatically in every plant studied so far. It is increasingly recognized that these microbes, as well as others including, e.g., epiphytes, viruses, ectomycorrhiza, N-fixers, etc., exert multiple effects on the plants particularly under adverse environmental conditions (Porras-Alfaro and Bayman, 2011; Hill et al., 2019; Dastogeer et al., 2020b; Kumar et al., 2020). These beneficial effects, which are strongly dependent on environmental conditions (Rodriguez et al., 2009; Hoeksema et al., 2010; Dastogeer, 2018), include priming against pests or herbivores, the acquisition of growth-limiting nutrients from the soil, tolerance to drought, salinity and thermal stress, etc. (Rodriguez et al., 2009; Goh et al., 2013; Hardoim et al., 2015; Azad and Kaminskyj, 2016; González-Teuber, 2016; Molina-Montenegro et al., 2016; Dastogeer and Wylie, 2017; Rho et al., 2018; Hill et al., 2019). Microbes produce a wide range of compounds to impact the responses of plants at the molecular level, triggering the biosynthesis of pigments, secondary metabolites, hormones, antioxidants, and alkaloids (Meena et al., 2017; Dastogeer et al., 2018; Basit et al., 2021; Naamala and Smith, 2021). Just as for other stresses, it is becoming increasingly evident that the heat stress responses of plants may be influenced by their interactions with microbes (Eerens et al., 1998; Ali et al., 2011; Khan A. L. et al., 2015; Ismail et al., 2020a; Li X. et al., 2021).

Despite the ubiquity and high diversity of microbial symbionts in plant tissues and the widespread effects and economic relevance of heat as a stress factor for plants, when compared with other abiotic factors such as drought, salinity, or nutrient limitation, the involvement of symbionts in plant host responses to heat stress have been received less attention (Meena et al., 2017; Kumar and Verma, 2018). The handful of studies available indicate that the magnitudes of the effect of microbes on plant heat stress tolerance and associated mechanisms such as accumulation of osmolytes, antioxidants, phytohormones, etc., vary among various studies (Meena et al., 2017; Kumar and Verma, 2018). We assume these differences could be associated with multiple factors such as level of stress, types of host and symbiont partners, environmental conditions, and their complex interactions. In order to gain insight into and harness the benefit of symbiont in agricultural management practice, it is imperative to elucidate the detailed mechanism of microbe-plant interaction under heat stress conditions (Harman and Uphoff, 2019). It is indeed tenuous and often erroneous to deduce the findings in general context from the individual investigation (Dastogeer, 2018). Therefore, to ascertain the central tendency and pinpoint the patterns of microbial influence on plants under stress and compare with them under control, it is paramount to integrate results across studies in order to determine if the factors can be known (Borenstein et al., 2009; Dastogeer, 2018). Here, we performed a meta-analysis to measure the overall strength and direction of effect (beneficial, harmful or neutral) of symbiosis on important plant characteristics associated with stress tolerance mechanisms. To the best of our knowledge, this aspect has not previously been discussed in a review of the published article, and thus present article fills a substantial gap in our understanding of interactions between plants and their symbiotic microbial partners.

A meta-analysis is a mathematical and statistical approach that pools the results of various investigations to estimate a mean effect size for treatment across a range of studies (Rosenberg et al., 2000). We can evaluate the findings of a study with respect to all other comparable studies to assess if the effect of a treatment is consistent across studies or if it differs substantially among studies and which factor might be accounted for the differences (Borenstein et al., 2009). The categorical variables or “moderators” are often included in meta-analyses to pinpoint which and how various features modulate the treatment effect of interest. This form of analysis has been, for example, used to determine the impact of microbial symbionts on plant response to salinity stress (Chandrasekaran et al., 2014, 2016; Rho et al., 2018), drought stress (Dastogeer, 2018; Rho et al., 2018), and cold stress (Acuna-Rodriguez et al., 2020). In the current study, we accumulated data from all studies to date and measured the effects of endophytes on 24 plant response parameters that encompass plant growth, photosynthesis, metabolites, and enzymatic activities that are subjected to change under thermal stress conditions. Our purpose was to answer the questions, broadly; what is the overall impact of microbial colonization on various plant growth and photosynthetic parameters of plants exposed to heat stress? Does the plant-symbiont relationship differ under heat-stressed conditions compared to unstressed conditions? What is the role of symbiosis in osmotic balance and antioxidant enzymes regulation in plants exposed to thermal stress? We supplemented our meta-analysis with a review of relevant literature for a deeper insight into this subject.

The meta-data were collected by following the general guidelines of Field and Gillett (2010). We performed a literature search in Web of Science (Clarivate Analysis) and PubMed database through September 2020. Our search terms were: endophyte or AMF or mycorrhiza or bacteria or fungi AND either “heat stress*” or “hot stress*” or “high temperature.” The Boolean truncation (“*”) was used so as to include the variations of the word such as fungi, fungus, and fungal. The search produced 4,562 unduplicated papers, and 90 peer-reviewed articles were considered likely to contain relevant information by reading the title and abstract. To finally select the papers for data extraction, we had preset criteria such as so that these studies can be included in our meta-analysis and extreme heterogeneity minimized (Cooper, 2015; Ahn and Kang, 2018; Cooper et al., 2019; Seidler et al., 2019):

Based on the above criteria, most articles were excluded for meta-analysis, and only 39 peer-reviewed original research articles were retained for the final analysis (Supplementary Table 1). We accepted studies to differ in the levels of fertilizer applied, growth conditions (greenhouse, growth chamber, or field), duration of stress application, the magnitude of stress, and growth media into our meta-analysis. Papers spanned 23 years (1998–2021) and were in English. We apologize for not including the potential paper that contains data written in other languages or those that we might have unintentionally missed during our search or due to stipulated selection criteria.

When a publication provided results of more than one study system/group, those systems were taken as independent data points (Hedges et al., 1999). From these articles, we extracted information on host identity (plant species, genus, family), symbionts’ identity (genus, species), plant biomass (shoot and root), length (shoot and root), relative water content, photosynthetic parameters, enzymes parameters, and other relevant data from the studies (Supplementary Table 2). The means, sample sizes (replications), standard deviation were recorded from each study. Standard deviation was calculated from standard errors (SE) using the equation: SE = SD (n^–1/2). Frequently the results were presented in a graph, and we used WebPlotDigitizer (Rohatgi, 2021) to digitize the values. If any paper reported multiple treatments or host/AMF combinations, these were included as an independent data unit in the analysis. However, it might increase the dependence of a particular study when considered them as an independent report (Gurevitch and Hedges, 1999). Nonetheless, this practice is assumed to increase the statistical power of meta-analysis (Lajeunesse and Forbes, 2003) and several biological meta-analyses papers also used the same (Holmgren et al., 2012; Veresoglou et al., 2012; Mayerhofer et al., 2013; McGrath and Lobell, 2013; Dastogeer, 2018).

Meta-analyses were conducted using the “metacont” function of the package “meta” (Balduzzi et al., 2019) implemented in R version 4.0.4 (R Core Team, 2021). Although the DerSimonian and Laird (DL) method is commonly used by default to estimate the between-study variance, it was suggested that for continuous data, the restricted maximum likelihood estimator (REML) are better alternatives to estimate the between-study variance (DerSimonian and Laird, 1986; Veroniki et al., 2016; Cooper et al., 2019). We used the Restricted maximum-likelihood estimator (REML) to make statistical inferences considering the effects as random. Standardized mean difference (SMD), which is defined as the ratio of the difference in mean performance between treatments to the pooled standard deviation, was calculated. This effect size measurement is done in meta-analyses that involve studies reporting continuous outcomes, similar to our study (Faraone, 2008). Generally, the SMD values of 0.3, 0.5, and 0.8 are considered as a small, medium, and large effect sizes, respectively (Cohen, 1988). Quantification of heterogeneity as well as testing the significance were estimated with Higgin’s I² and Cochran’s Q statistics. “I²” is defined as the ratio of true heterogeneity to total heterogeneity across the effect sizes, whereas “Q” is the weighted deviations of the summary effect size that is due to true heterogeneity rather than sampling error (Higgins and Thompson, 2002; Higgins et al., 2003; Huedo-Medina et al., 2006). The I² values < 25%, 25–75%, and > 75% are regarded as representing low, moderate, and high heterogeneity (Higgins et al., 2003) by convention. The coefficient of interval estimate of effect sizes was set to 95%, and the effect size (SMD) was considered significant when 95% CIs (confidence interval) did not cross zero. SMD of zero suggests that the two treatments (microbial inoculated or non-inoculated) have equivalent effects; SMDs greater than zero indicate the degree to which the microbial inoculated outperforms the non-inoculated samples and vice-versa.

Publication bias for each dataset of different parameters was tested by inspecting funnel plots and Egger regression test for funnel plot asymmetry (Begg and Mazumdar, 1994; Egger et al., 1997; Simonsohn et al., 2014). We found that for most of the parameters, there were no substantial publication biases in these datasets (Supplementary Table 2). For some parameters which showed some biases, then we took the effect sizes (SMD), CIs, and heterogeneity statistics after applying the trim-and-fill method to correct the biases (Duval and Tweedie, 2000). However, for the majority of the cases, the trim and fill did not substantially alter (decreased) the SMD values than the untrimmed SMD values (Supplementary Table 2). Therefore, we used the untrimmed SMD values (i.e., did not adjust for publication bias) for the sake of consistency for all the parameters for creating forest plots. After carefully observing the heterogeneity statistics as well as significance level for SMDs, we performed subgroup analyses for root length parameters to determine the influence of the factors such as plant or endophyte identity, type, etc., because sufficient data were available. For a factor to be included in the analysis as a subgroup variable, it had to be reported from at least four studies.

The articles we reviewed presented findings of experiments involving various host-microbe systems. These investigations considered various parameters in evaluating the influences of microbial colonization in providing heat stress tolerances of plants such as plant biomass and response at the physiological, molecular, metabolic as well as the hormonal level (Figure 1). Effect of fungi was encountered more often, which accounted for 69% of the 42 articles used in our meta-analysis. In general, both fungal and bacterial symbionts were found to increase plant fitness under thermal stress by augmenting photosynthesis, improving antioxidant responses of plants, triggering earlier hormonal signaling, enhancing osmolyte and nutritional balance (Table 1). In the next section, we outlined the influence of microbial symbionts on plant thermal stress tolerance as obtained from our meta-analysis and discussed in light of available literature.

There is a huge lack of understanding of the mechanisms of microbe-mediated plant heat stress tolerance at the molecular level. The heat shock protein (HSPs) act as “molecular chaperones,” and their upregulation is well known to increase thermotolerance in plants. The expression of heat stress-dependent regulatory proteins such as heat shock transcription factors (HSFs) and heat shock proteins (HSPs) have been described. The HSPs are crucial in regulating and restoring protein structures and maintaining the condition of plants under heat stress (Queitsch et al., 2000; Nakamoto and Vigh, 2007; Xu et al., 2012; Wang et al., 2014). Several studies reported an increased level of expression of HSPs in the microbial colonized plant compared to non-colonized plants under heat stress (Abd El-Daim et al., 2019; Bruno et al., 2020; Li X. et al., 2021). To reveal the mechanisms through which Enterobacter sp. SA187 mediates plant thermotolerance in Arabidopsis Shekhawat et al. (2020) carried out RNA-seq and targeted transcriptomic analyses. They described that the bacteria-mediated heat tolerance is associated with the gene networks for heat protection of plants similar to that of the heat priming (pre-exposure of plants to heat that enables plants to be better prepared to response to heat stress). For example, heat tolerance is associated with the upregulation of genes related to the auxin and gibberellin hormones and the down-regulation of the genes related to jasmonic and abscisic acid. It was also found to be related to hyper-induction of heat-responsive and heat memory genes, which is proven in heat primed plants (Lämke et al., 2016; Ling et al., 2018; Liu et al., 2018). The bacteria inoculated plants resulted in higher expression of several heat-responsive genes such as HSFA2, HSP101, HSP70, HSP70B, GA3OX1, HSP90.1, and ATERDJ3A as well as heat memory genes; APX2 (ascorbate peroxidase 2), HSP18.2, H3K4me3 (epigenetic modification to the DNA packaging protein Histone H3) (Lämke et al., 2016; Sedaghatmehr et al., 2016). In a study, increased expression of heat shock protein gene HSP17.8 and Dehydration-responsive gene DREB1 and decreased expression of POD47 and FeSOD have been reported for fungal (Aspergillus aculeatus) conferred heat stress tolerance in Lolium perenne grass (Li X. et al., 2021). Khan et al. (2020b) reported that GmLAX3 and GmAKT2, the key genes involved in the regulation of the ABA-dependent pathway, were downregulated initially but upregulated after 5 and 10 days in the inoculated soybean plant under thermal stress. The expression of these genes may be linked with plant stress response by modulating auxin and ABA signaling, decreasing ROS production and enhancing K⁺ gradients, which are critical in plant tissues under various stresses (Zhou et al., 2014; Chai et al., 2016). The B. velezensis inoculation induced an increased abiotic stress tolerance, and the pattern of metabolic modulation and abundance of several proteins were found to be similar in wheat challenged with different abiotic stress factors such as heat, cold, and drought (Abd El-Daim et al., 2019).

Taken together, high-temperature influences plant growth by affecting photosynthesis, osmotic balance, enzymatic activities, and metabolic activities. Microbial symbionts consistently help plants reduce the impact of thermal stress by manipulating physiological processes (Figure 7). Microbe-induced plant heat stress tolerance has broad ecological and agricultural implications. In some regions, increased stress tolerance can result in higher crop yield under higher temperature environments. Continued interest in the microbiome and their functional roles in uncovering the underlying mechanisms of plant-microbe interaction appears well-justified.

Despite the evidence of global warming and the importance of heat stress on plant ecology and productivity (Wahid et al., 2007; Teixeira et al., 2013; Jagadish et al., 2021) and the increasing evidence of microbial functions of plant stress tolerance (Table 1 and Figure 1), the current review here identified a significant lack in our knowledge on microbial conferred plant performance and productivity under high-temperature conditions. Apparently, the most crucial of these is the lack of experiments that demonstrate how do microbes actually influence plant responses in the natural environment. Most of the investigations reviewed here were carried out under controlled environmental conditions and on cultivated host species. Therefore, experiments focusing on the impacts of symbionts on wild and crop plants exposed to heat stresses in the natural environment are required. In a recent paper, Ballesteros et al. (2020) analyzed metatranscriptomic data from the Antarctic plant Colobanthus quitensis, highlighting that the plant-associated fungal community provides beneficial effects, such as increased plant growth, tolerance to abiotic stress, and higher leaf carbon gain, when plants are grown under a simulated global warming scenario. The reports from these investigations should include mean values of response variables, measures of variance, and replication and preferably provide the actual dataset as the Supplementary Data so that these data can be used in future meta-analyses. Although it was shown that increasing temperatures due to climate change might negatively impact the outcomes of symbioses (Kiers et al., 2010), the current meta-analyses indicated that microbial symbionts have more positive effects on plant performance under heat stress than under normal conditions. It is not clear why microbial inoculation exerts its effects in more stressed conditions than under normal temperatures conditions, but putatively, this is associated with plant stress acclimation at the molecular or cellular levels. Tolerance of non-colonized plants to heat stress is well studied, and the reports suggested that the mechanisms are associated with enhanced expression of a variety of heat shock proteins, other related proteins, production of ROS, and induction of mitogen-activated protein kinase (MAPK) and calcium-dependent protein kinase (CDPK) cascades, and, most importantly, chaperone signaling and transcriptional activation (Wahid et al., 2007). As previously shown for abiotic stresses such as freezing and drought and assumed for microbial conferred cold adaptation (Knight and Knight, 2001; Acuna-Rodriguez et al., 2020), some kind of cross-talk between the proteins common to signaling pathways are present for responses to heat and microbial colonization leading to enhanced plant metabolism. The involvement of microbial elicitors at the plant cell surface resulting in increases in cytosolic Ca2⁺ concentrations could be another possibility (Sanders et al., 1999; Acuna-Rodriguez et al., 2020). However, more research is needed to establish whether these or similar mechanisms might explain the beneficial effects of microbial symbionts on plants at higher temperatures. As discussed by Acuna-Rodriguez et al. (2020), less attention has been paid toward understanding the differences in the optimum growth temperatures of microbial symbionts and their plant hosts. For instance, in the Epichloe symbiosis, the seemingly decreased plant performance at low temperatures in the natural conditions could be that the minimum temperatures requirement of the endophytes being higher than those of their hosts (Casler and Van Santen, 2008; Wäli et al., 2008). Extensive surveys to identify novel microbial symbionts capable of improving plant performance under stress are advocated. Isolation of naturally occurring plant and soil microbiomes and screening them for potential symbiotic benefits is a continuous process. Developing a rapid but robust screening method is important so that a large collection of microbes can be tested for their effects, and the results should be supported by field experiments in order to translate the research into the field. A few surveys focused on endophytes collected from hot environments, and some of the endophytes tested were found to grow at high temperatures (Zhou et al., 2015; Dastogeer et al., 2019). Studies on other components of the plant microbiome such as algae, protists, and viruses revealed that they are also present in the tissues or in the rhizosphere of many plant species and provide some fitness benefits (Roossinck, 2015; Gao et al., 2019; Dastogeer et al., 2020a; Dumack and Bonkowski, 2021; Lee and Ryu, 2021). For example, rhizosphere protists altered the metabolites of the maize plant, particularly, polyols. This alteration is presumably to be associated with reduction of the impact of stress caused due to grazing (Kuppardt et al., 2018). A plant virus, yellowtail flower mild mottle virus (YTMMV, genus Tobamovirus), enhanced drought stress tolerance in Nicotiana benthamiana plants by modulating its metabolic compounds (Dastogeer et al., 2018). However, the effect of many of the microbiome components has received less attention. We also suggest studies that consider microbial symbionts on plant heat tolerance should take into account the involvement of secondary metabolites because many microbes utilize these compounds in the interaction with plants and with environments (Zaynab et al., 2018; Ullah et al., 2020). Another important area of research, but has so far been neglected, is to study the multi-partite interaction of microbes and their impact on stress tolerance, such as thermal tolerance. It was found that a fungus can increase the heat tolerance of a plant only in the presence of a virus within it, whereas neither virus nor the plant separately could confer this benefit. Therefore, a three-way symbiosis is required for thermal tolerance. In nature, plants do not remain in isolation but in continuous interaction with a wide variety of microbes (Marquez et al., 2007). In addition to microbe-plant interaction, microbe-microbe and microbe-environment interaction should be taken into account to understand the stable functions of a particular microbe in its potential application in agricultural settings, which is a dynamic environment and subject to continuous changes (Dastogeer et al., 2020a).

KD developed the concept, performed the database search, partially collected the data, analyzed and interpreted the results, and wrote the draft. MZ, MR, and MS extracted the data, produced the graphs, discussed the results, and revised the manuscript. AC produced the table and discussed the results. All authors approved the final version of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.