Risk and reward of the global truffle sector under predicted climate change

To date, no global database of TGRs has been compiled, primarily because of the secrecy of the location of truffle plantations, which reflects the reluctance of sharing specific information by truffle farmers and hunters. In order to define the global distribution of the main Périgord TGRs that substantially contribute to international markets, we reviewed peer-reviewed scientific articles, monographs, technical notes and conference proceedings (table S1 available online at stacks.iop.org/ERL/17/024001/mmedia). All information was published after 2010, either in indexed databases or on the truffle growers' association websites. Of the dozens of relevant publications that included the search words 'truffle' or 'truffle plantation' or 'truffle cultivation' in conjunction with toponyms, only ten articles contained useful information about specific TGRs. While we restricted our focus to those regions in which production has been reported (table S1), we excluded all regions with established plantations lacking any recent harvest (e.g. China; C. Colinas, pers. comm.) or showing globally insignificant production rates (e.g. Morocco, Israel, the United Kingdom, eastern United States). Finally, we neglected occasional individual truffle findings and theoretically suitable areas for truffle growth (Thomas and Büntgen 2017), such as the region between Texas and Virginia (Lefevre 2012).

To delineate and digitize all TGRs, we used the QGIS ver. 3.16 (QGIS.org 2022) to produce a shapefile in which polygon features represent the TGRs. In Europe (Spain, Italy and France), Chile, South Africa and New Zealand, we defined boundaries based on 1st-level administrative subdivisions (ASs). In the United States where this administrative scale corresponds with individual states, we used 2nd-level AS instead. Shapefile data on both spatial levels are available at the 'Database of Global Administrative Areas' (see SI). While the 1st-level AS in Australia corresponds to large territories and the 2nd-level AS (small local government bodies) is too small for resolution of GCMs, we defined TGRs by temperate and Mediterranean climate zone according to Köppen. These groups sufficiently cover the spatial distribution of currently operating truffle plantations (table S1) and are considered suitable for truffle cultivation (Malajczuk and Amaranthus 2007). Data on climate zones are freely available at Australian Bureau of Meteorology (see SI). To clarify the relationship between temperate (all subtypes: Cs, Cw, Cf) and Mediterranean climate (only Cs), we provided the corresponding subtypes to each TGR (Beck et al 2018) (table S2). Data for the climate diagrams based on instrumental records ('present climate') are available at National Oceanic and Atmospheric Administration (NOAA) (see SI).

Périgord truffles grow at sites with an average annual temperature of around 12 °C and precipitation totals of approximately 780 mm. The optima for the warmest and coldest months (July and January) are 20.5 °C and 3.8 °C, respectively. Summer precipitation totals must not fall below 140 mm (Čejka et al 2020). Although Périgord truffles are harvested in winter from around November to March (in the northern hemisphere), fruitbody formation likely starts as early as the previous spring, after mycorrhizal colonization of the host tree roots and mycelium formation (Garcia-Barreda and Camarero 2020). Between spring and late summer, the growth and maturation (peridium and gleba development) continue. This process mainly depends on the amount of water in the soil and host plants, although excess precipitation in the cooler autumn season may affect truffle harvests negatively (Büntgen et al 2019). By the same token, the increased frequency of climatic extremes, such as drought spells and sub-zero temperatures, often impair future harvests. In addition to reduced truffle yields, prolonged summer drought can affect fruitbody maturation, while winter frosts may cause peridium rupture, especially if the truffle contains large amounts of water (Thomas and Büntgen 2019).

Since growth and ripening of Périgord truffles, which are mostly found in the topsoil layer, depend on natural summer rainfall and temperature (Büntgen et al 2019), we selected three modeled climate variables for the summer season: temperature (Kelvin), precipitation flux (kg m⁻² s⁻¹), and moisture in the upper ∼10 cm of the soil column (kg m⁻²). This last variable was preferred over column-wide moisture because GCMs vary in soil depth. That, in addition to the non-uniform vertical distribution of soil water, makes GCMs not comparable (Berg et al 2016). While climate estimates of temperature and precipitation were derived from a set of 20 GCMs, soil moisture was limited by the availability of only 17 models (table S3). All GCMs used in this study contributed to Coupled Model Intercomparison Project – Phase 5 (CMIP5) as part of the Intergovernmental Panel on Climate Change (IPCC) Assessment Report 5 (Flato et al 2013), and are available from the World Data Center for Climate (see SI). Our preference for GCMs over regional climate models is described in the online SI. To estimate a wide range of climate change impacts on TGRs throughout the 21st century (2006–2100 CE), we used both a 'moderate' and 'severe' climate scenario represented by the Representative Concentration Pathways (RCP)_4.5 and RCP_8.5, respectively. In ensuring consistency and avoiding over-representing the forced responses of GCMs that submitted more realizations to the CMIP5 dataset, we employed only the first ensemble member from each model (r1i1p1). To process the GCMs, as well as any other climate data (NetCDF format), we used the Climate Data Operator (CDO) 2.0.0 software (Schulzweida 2021). For a detailed description of the processing and bias correction of the data using CDO, see the online SI.

Using the QGIS ver. 3.16 (QGIS.org 2022) we exported vertices' coordinates of all shapefile's polygons into a delimited text file, which was then imported into the CDO to extract future projections of climate variables in TGRs. To remove oceanic regions from all variables, we applied a similar procedure to the global landmass shapefile (see SI) and created a mask over which we computed the area average, which we hereafter considered to be the global mean. Since the Climatic Research Unit (CRU) and Global Land Data Assimilation System (GLDAS-2.1) baseline datasets do not spatially cover high-latitudes in the Southern Hemisphere, preventing local bias-correction, Antarctica was excluded from the global landmass. When calculating a mean of each timestep (summer season) for TGRs and the globe, we accounted for grid-specific weights derived from the cell area of the Gaussian grid. Finally, with summer data being related to June–August (JJA) and December–February (DJF) in the Northern and Southern Hemispheres, respectively, we computed estimates of seasonal means from monthly temperature and soil moisture data, and seasonal sums from daily precipitation totals/fluxes (see SI). Since the austral summer data for 2100 terminate by the end of the year, we excluded 2100 from our analysis.

We examined the direct response of the truffle sector in each TGR to projected climate change using the impact analysis. We estimated Périgord truffle price levels for each TGR for the late 21st century based on historical data from the world's two most important truffle exporting countries in Europe (see Statistical analysis and modeling). We used historical prices from wholesale markets in France (1995–2020/2021) provided by Agrimer, France (https://rnm.franceagrimer.fr), and Spain (1997/1998–2020/2021) compiled in 2021 from multiple sources including Reyna (2012), Oliach et al (2020b) and Teruel Truffle Growers Association (https://trufadeteruel.com/atruter/). Similarly, we estimated Périgord truffle production rates in each TGR for the late 21st century based on historical records from Spain, France, and Italy (Groupement Européen Truffe et Trufficulture; Spanish Federation of Truffle Growers Associations; French Federation of Truffle Growers; Oliach et al 2021). Therefore, we were also able to provide estimates of future price levels and production rates for each TGR (see Statistical analysis and modeling and Results). However, we did not include historical records from non-traditional TGRs, because local truffle cultivation has only recently begun in some regions. That, in addition to exponential truffle growth reflecting the number of plantations rather than natural variability in precipitation, makes reliable predictions difficult. Estimating future parameters in non-traditional TGRs is further problematic due to the lack of official and homogenized records. To the best of our knowledge, the only existing truffle harvest data are limited to personal inventories and biased towards a small number of plantations.

To detect long-term trends in the climate projections, we applied a non-seasonal Mann–Kendall test on the bias-corrected 2006–2099 anomalies relative to the 2006–2020 baseline mean (Mann 1945). This non-parametric statistical test is commonly used to detect increasing or decreasing trends over time for many types of data, including groundwater, water quality, streamflow, lake levels, temperature, and precipitation. In terms of functionality, the Mann–Kendall statistics were determined by the ranks of the original values and do not require homoscedasticity and normal distribution (Wang et al 2020).

Moreover, we observed if climate parameters are likely to get significantly more or less variable until the end of 21st century by testing heteroscedasticity of the bias-corrected area-average of TGRs and the remaining global land surface between 2010 and 2039 and the distant future from 2070 to 2099 CE. We used prior-ANOVA non-parametric Levene's test (Levene 1960; table S4), after inspecting (non-)normality via the Shapiro–Wilk test (Shapiro and Wilk 1965; table S5). We used the Shapiro–Wilk test because it is not as affected by ties as the Anderson–Darling test, and we chose Levene's test over Bartlett test because it is less sensitive to departures from the mean. To understand spatial climatic extremes, we developed plots showing the distribution of standard deviation differences of annual variability between the early (2010–2039) and late (2070–2099) period for each grid cell in each TGR.

Based on a strong dependency of winter Périgord truffle production on previous summer precipitation (p < 0.00001; Büntgen et al 2019), and the adverse effect of heatwaves on truffle growth (Garcia-Barreda and Camarero 2020), we developed multiple linear regression (MLR) models with two independent variables. These were represented by summer precipitation and the heat index of the warmest months (July and January). The latter is a physiological parameter combing the effect of air temperature and relative humidity. However, since the heat index is not usually measured, we used the July/January temperature and relative humidity to calculate this parameter according to the conventional equation (see SI). Historical and projected relative humidity data were obtained from European Centre for Medium-Range Weather Forecasts Re-analysis (ERA5) and World Data Center for Climate (GCMs), respectively (see SI). Using a common period of 24 years from 1997/1998 to 2020/2021, we developed five MLR models (tables S6–10) between historical climate (CRU TS4.05 or ERA5) and national price levels (Spain and France) and production rates (including Italy). We applied three models for production rates and two models for price levels on bias-corrected future climate data to estimate these parameters in each TGR for a three successive 24 year periods: 1997–2020, 2021–2044 and 2076–2099. Both independent variables were used in all five MLR models except for the association between Spanish production and the heat index, which lacked correlation (r = −0.01) and showed multicollinearity. We then used the sum of the three and the average of the two sets of modeled production and price values, respectively. We did not standardize any of the data used.

A subtropical high-pressure ridge between spring and autumn forms the Mediterranean climate of all TGRs, and most of the annual precipitation supply depends on mid-latitude westerly winds in winter. With a warming-induced strengthening of the mid-latitude jet stream and a reduction of land–sea temperature gradients that maintain subtropical anticyclonic blocking and prolong summer droughts (Tuel and Eltahir 2020), the winters are likely to shorten, and precipitation totals are expected to decline in many TGRs (Cramer et al 2018). Exceeding previous calculations by about ∼15%–20% (Jacob et al 2014), JJA precipitation totals in southern Europe are predicted to decrease by almost ∼30%. That, in addition to a predicted increase of JJA temperatures ∼20% faster than the global average (figure 2 and Cramer et al 2018), will significantly accelerate soil drying by the end of the 21st century. Similar stressors will exacerbate future drought in the United States' TGRs (Mann and Gleick 2015), and increasing temperatures, decreasing precipitation, and soil aridification in central Chile will likely lead to ∼40% drier summer conditions from DJF by the end of the 21st century (Araya‐Osses et al 2020).

Future summer warming is projected for the whole of South Africa, although hydroclimatic conditions differ between the west and the east of the country (Roffe et al 2019). At the western coast (Western Cape), where the climate is usually drier (Roffe et al 2019), soil moisture content and precipitation totals in the austral summer and winter are expected to decrease over the 21st century (figure 2 and Pascale et al 2020). This, as well as the observed decline in winter precipitation across eastern South Africa in the past century (Net 2009), is in principle due to the drying and shifting of subtropical high-pressure ridge towards the pole (Rojas et al 2019, Pascale et al 2020). In contrast, summer precipitation in eastern South Africa (KwaZulu-Natal) exhibits a positive trend under the influence of the 'warm' Agulhas current bringing more air moisture (Beal et al 2011; figure 2). This is comparable with downscaled climate projections indicating increasing summer precipitation in parts of east-central South Africa (Hewitson and Crane 2006, Almazroui et al 2020), probably caused by intensified short-term heavy precipitation (Dosio et al 2019).

With future warming, high-emission projection for summer precipitation based on our GCMs' ensemble mean reveal non-significant trend for western Australia over the 21st century (figure 2), probably because of low agreement between the existing GCMs (Raut et al 2017). Some of these climate models, however, project a slight increase in summer precipitation in eastern Australia, which is consistent with the ensemble mean used in this study (figure 2 and Timbal et al 2017). In contrast, comparison between historical (1986–2005 CE) and future (2080–2099 CE) periods suggests that summer precipitation totals will increase by only 1% (Timbal et al 2017). Positive trends in summer precipitation have been suggested to originate from the intensification of short-term heavy rainfall events at timescales of hours to days (Dey et al 2019). Either way, decreasing water content in soil during summer in both Australian TGRs (RCP_4.5; figure 2) corresponds with an annual decline in precipitation, particularly in winter (JJA), when most precipitation occurs (Delworth and Zeng 2014, Dey et al 2019). A decrease of winter precipitation is also projected for New Zealand's TGRs in the lowland east, as opposed to those plantations at higher elevations in the west, where precipitation is predicted to increase. However, summer precipitation under high-emission scenario is projected to increase in New Zealand, but decrease in the western part of this country. Fluctuating projected precipitation across New Zealand and a projected stability of soil moisture content are associated with a decreasing pace of westerly winds (Mullan et al 2018), and an increase in the frequency rather than the intensity of precipitation (Sansom and Renwick 2007).

Future warming will increase the frequency and intensity of extreme summer temperatures in most of the TGRs (figure 3). Comparison of measured and predicted climate data from Europe (1961–1990 versus 2071–2100 CE) shows that heatwaves are likely to become twice to five times as long, and their frequency will increase from one every three to five summers to about two to three heatwaves per season (Fischer and Schär 2010). In the United States, the frequency of heatwaves associated with longer drought episodes will double after 2050 CE, compared to the earlier period 1950–2000 (Cook et al 2015). In the Southern Hemisphere, increasing extremes of summer temperature and soil moisture content associated with intensified El Niño Southern Oscillation (ENSO) in the 21st century will cause more frequent heatwaves and longer drought spells in central Chile (Cai et al 2020), South Africa (Pascale et al 2020), and western Australia (Andrys et al 2017). Even where the influence of ENSO is relatively minor, extreme summer temperatures in eastern Australia are expected to rise (King et al 2017), and heatwaves in late-21st century are predicted to be common for New Zealand, likely due to low wind speeds associated with unusually positive phases of the South Annual Mode (Salinger et al 2019). While a more detailed discussion of possible drivers behind projected climate conditions is beyond the scope of this study, the uncertainty associated with the GCMs appears crucial for further research (see our expansion on GCMs uncertainty in SI).

As with other fungal species (Kauserud et al 2012, Diez et al 2013, Karavani et al 2018), anthropogenic climate change will exert regionally disproportionate effects on the quantity and quality of truffle yields in the 21st century. With the warm period largely stimulating truffle maturation, the subsequent winter harvest depends on the amount of the preceding summer precipitation (Büntgen et al 2019). Therefore, rapid summer warming, and a long-term precipitation decline in southern Europe will reduce the local truffle winter harvest by 39% and in turn increase prices by 66% (figure 4). Although there will be a 56% and 81% decline in truffle production in the western United States and Chile, our MLR models project a price change of only +8% and −7%, most likely due to slower warming. Disparity between production rates and the price levels is related to the fact that estimates of both variables depend only on climate, while lacking any economic behavior. Moreover, estimates in our models cannot reflect daily climate variability and short-term tolerance of truffles to heat stress (Garcia-Barreda et al 2019) via drought-resistant strains (Mello et al 2006). Due to the opposing relationship between October and November precipitation totals and Mediterranean truffle winter yields (Büntgen et al 2019), we suggest that the harvest in some TGRs, including those in the Southern Hemisphere (Rojas et al 2019), could be enhanced by a future decrease of precipitation in late autumn/early winter. This would be 'good news' for truffle industries in eastern Australia and New Zealand, where a weaker rate of evaporation from soil, coupled with positive precipitation anomalies are likely to mitigate shrinking water resources. That, combined with slower temperature increase in summer will maintain winter truffle production at a level comparable to 1997–2020 (+9.5% and −0.2%). With a price increase (decrease) of only 19% (75%), eastern Australia (New Zealand) is likely to provide more affordable truffles (figure 4).

All TGRs will require intensive irrigation to mitigate the negative effects of the predicted trends and extremes of anthropogenic climate change. Although current irrigation rates are estimated to buffer some future moisture losses in some TGRs (Cook et al 2020), sustaining such rates will likely not be possible under the gradual warming-induced decline in (sub)surface and ground water reserves (Trnka et al 2018). This, added to enhanced water conflict from other socio-economic factors and the expansion of plantations associated with changes in agricultural land-use for other crops will challenge the additional water supplies, particularly in the fast-warming TGRs in Europe, the United States, Chile and western South Africa, where summer precipitation is projected to decline until 2100 CE. However, a surplus of water from irrigation, not based on adequate knowledge of the truffle's lifecycle combined with ineffective cultivation techniques, might reduce the harvest through enhanced microbial decay (Eslick 2013). Furthermore, even without high-intensity irrigation, spatially variable extreme precipitation events in a warmer climate (figure 3) could super-saturate plantation soils, causing surface erosion, reduced soil oxygen content, and alternation of the physiochemical properties of soil associated with a decrease in the diversity and abundance of truffles (Barnes et al 2018, Yan et al 2019). Therefore, the extent, frequency and intensity of irrigation in truffle plantations should not only compensate for future summer precipitation deficits (e.g. estimated ∼430 m³ ha⁻¹ in Europe under RCP_8.5), but also incorporate understanding of edaphic factors and ecophysiological adaptations of associated host partners, such as drought-controlled rhizogenesis and hydraulic redistribution (Büntgen et al 2019). We emphasize the necessity of increasing knowledge of fungus-host interaction under future ecological range shifts and phenological widening (Kauserud et al 2012), as well as for changes in resource availability. Nutrients for mycelial development, which starts earlier under warmer conditions, are depleted over a longer period, limiting subsequent fruiting body growth, and reducing fungal abundance and diversity (Gange et al 2013, Damialis et al 2015). Moreover, truffle growers should account for potential indirect effects associated with pathogens (Leiodes cinnamomea) (Büntgen et al 2019), especially for the cultivation of truffles in the conditions of rapid warming. On the other hand, adaptation of the truffle industry should take advantage of existing measures, such as the use of drought-tolerant host species. The sector should further benefit from elevational shifts in suitable climatic conditions that have already been observed in other fungal species (Diez et al 2020).

In addition to implications for agriculture and ecology, climate-induced changes of the global truffle sector will have wide-ranging economic consequences (figure 5). While the truffle sector is vulnerable to global challenges such as the financial crash in 2007/2008 or the current COVID-19 pandemic that resulted in both lower demands due to closures of businesses, and harvest declines associated with less workforce (Büntgen et al 2021), region-specific future climate developments are likely to lead to large-scale shifts in global truffle production and trade. In particular, the stability of the truffle sector will worsen in fast-warming regions with an estimated precipitation and thus production decline such as in southern Europe, where plantations expand each year by ∼2220 ha and harvesting employs over ∼210 000 people together with an even higher number of truffle-searching dogs (Reyna and Garcia-Barreda 2014). Conversely, expanding TGRs such as those in eastern Australia and New Zealand will profit from increased (or at least stable) production, while new suitable regions in Turkey and the greater Caucasus region may benefit from starting cultivation interest, though further educational support is required to encourage development in areas without a history of truffle culture.

Zoom In Zoom Out Reset image size

With increased human involvement in cultivation and eco- or myco-tourism, the prestige and exclusivity of the truffle sector is expected to rise, regardless of the impacts of climate change. This trend, however, requires careful management, as unlawful harvesting and trading, together with the introduction of competitive truffle species and the reduction of genetic diversity, could become serious threats to sustainable truffle production and product quality protection, and will need robust legal regulation in both established and emerging TGRs worldwide. With regional support being pivotal in coping and capitalizing on the predicted changes yet to come, our findings offer a tool to inform future policy, investment and decision-making in the development sustainable, optimized, global cultivation truffle management.