Global Change Biology (2015) 21, 3499–3510, doi: 10.1111/gcb.12960
Increased evapotranspiration demand in a Mediterranean climate might cause a decline in fungal yields under global warming R E D A 1 , 2 , B E A T R I Z AG UEDA2, JOSE M. OLANO2,3, SERGIO M. T E R E S A AG 4 5 V I C E N T E - S E R R A N O and M A R I N A F E R N AN D E Z - T O I R AN 2 1 Genius Medioambiente, C/.Campaneros 26, 42200 Almazan, Soria, Spain, Area de Botanica, Departamento de Ciencias Agroforestales, EU de Ingenierıas Agrarias, Universidad de Valladolid, Campus Duques de Soria, 42004 Soria, Spain, 3Sustainable Forest Management Research Institute, Universidad de Valladolid and INIA, Avda. de Madrid 44, 34004 Palencia, Spain, 4 ~ana 1005, 50059 Instituto Pirenaico de Ecologıa, Consejo Superior de Investigaciones Cientıficas (IPE-CSIC), Avda. Montan Zaragoza, Spain, 5Departamento de Produccion Vegetal y Recursos Forestales, EU de Ingenierıas Agrarias, Universidad de Valladolid, Campus Duques de Soria, 42004 Soria, Spain
Abstract Wild fungi play a critical role in forest ecosystems, and its recollection is a relevant economic activity. Understanding fungal response to climate is necessary in order to predict future fungal production in Mediterranean forests under climate change scenarios. We used a 15-year data set to model the relationship between climate and epigeous fungal abundance and productivity, for mycorrhizal and saprotrophic guilds in a Mediterranean pine forest. The obtained models were used to predict fungal productivity for the 2021–2080 period by means of regional climate change models. Simple models based on early spring temperature and summer–autumn rainfall could provide accurate estimates for fungal abundance and productivity. Models including rainfall and climatic water balance showed similar results and explanatory power for the analyzed 15-year period. However, their predictions for the 2021–2080 period diverged. Rainfall-based models predicted a maintenance of fungal yield, whereas water balance-based models predicted a steady decrease of fungal productivity under a global warming scenario. Under Mediterranean conditions fungi responded to weather conditions in two distinct periods: early spring and late summer–autumn, suggesting a bimodal pattern of growth. Saprotrophic and mycorrhizal fungi showed differences in the climatic control. Increased atmospheric evaporative demand due to global warming might lead to a drop in fungal yields during the 21st century. Keywords: bimodal pattern, climate change, global warming, long-term monitoring, mushrooms production, mycorrhizal fungi, phenology, saprotrophic fungi Received 6 March 2015 and accepted 10 April 2015
Introduction Fungi play a critical role in forest ecosystems, decomposing organic matter and turning it into inorganic components accessible to tree roots. Moreover, mycorrhizal fungi form symbiotic associations with tree roots, improving trees’ water and nutrient availability while providing defense against pathogens, thus boosting tree growth and survival rates. Beyond the provision of these basic ecosystem services, mushrooms are also a valuable source of wild food. Mushroom-picking is the basis for a large economic sector including mushroom transformation and commercialization, and even cre ates important benefits for the tourist sector (Agreda et al., 2014).
Correspondence: Jose M. Olano, tel. +34 975 129 485, fax +34 975 129 201, e-mail: [email protected]
© 2015 John Wiley & Sons Ltd
A sustainable management of wild mushrooms requires a thorough comprehension of the factors driving mushroom production and temporal variability. Control of mushroom production is related to multiple environmental factors such as site soil characteristics (Barroetave~ na et al., 2008; Alonso Ponce et al., 2014), endogenous dynamics of fungal communities (Smith et al., 2002; De Miguel et al., 2014), succession (Fern andez-Toiran et al., 2006; Baptista et al., 2010), forest management (Egli et al., 2006; Le Tacon et al., 2014), and climate (Baptista et al., 2010; Diez et al., 2013). Climate is a major component determining mushroom production, particularly during the fruiting period (Boddy et al., 2014). Increases in temperature and modification of rainfall patterns may have already altered mushroom phenology (Kauserud et al., 2008; B€ untgen et al., 2013b; Diez et al., 2013; Gange et al., 2013), as well as mushroom productivity (B€ untgen et al., 2012b), with positive and negative outcomes being detected 3499
R E D A et al. 3500 T . AG depending on the study region (B€ untgen et al., 2012a; Sato et al., 2012). Most fungi produce ephemeral fruiting bodies that can be observed only for a few days each year, making the acquisition of accurate annual productivity data time-consuming and challenging. In fact, progress in unveiling the climatic forcing of fungal productivity has been hindered because of the scarcity of long-term time series based on systematic sampling (Zambonelli et al., 2012; Boddy et al., 2014). Indirect estimates based on market prices of commercial fungi have been used to estimate abundance and relate it to climate (B€ untgen et al., 2012a; Le Tacon et al., 2014), while national inventories based on collaborative work also provide some information on fungal relative abundance and phenology (Diez et al., 2013; Boddy et al., 2014). However, the value of these approaches is often limited by inconsistency in the sampling effort or design; more accurate prediction requires high-quality time series data sets based on the systematic sampling of fungal productivity. However, existing time series mostly comprise relatively short periods (Salerni et al., 2002; Bonet et al., 2008), with longer records being scarce (Egli et al., 2006; Martınez-Pe~ na et al., 2012; Agreda et al., 2014) and missing some biogeographical areas such as the Mediterranean. Climate warming has been strong in the past five decades in the western Mediterranean region (Alpert et al., 2008; Bertin, 2008) with a range of increase between 0.2° and 0.4 °C decade1, which has dramatically increased the atmospheric evaporative demand (AED) (Vicente-Serrano et al., 2014a). Current climate change models predict an intense temperature increase for the coming decades that ranges from 2 to 3 °C for the 2040–2070 period (Giorgi, 2006; GarcıaRuiz et al., 2011) that would cause higher AED and a general decrease of water resources (Garcıa-Ruiz et al., 2011; Vicente-Serrano et al., 2014b). Although the projections for precipitation are less reliable than for temperature (Giorgi et al., 2004; Deque et al., 2005), most studies project a general tendency toward less precipitation in the next century (Giorgi & Lionello, 2008; Evans, 2009). The expected higher AED and lower precipitation mean that more frequent and severe drought events are therefore predicted (Dai, 2013; Cook et al., 2014). A decrease in water availability during the growing season would severely impact Mediterranean forest ecosystems as it would reduce tree growth and increase tree mortality (Carnicer et al., 2011; Pasho et al., 2011). Mediterranean mushrooms may also be affected by increasing aridity; directly so, because fungal growth and sporocarp production is closely linked to soil humidity (Murat et al., 2008), but also indirectly because reductions in
primary productivity will limit resource availability to saprotrophic fungi, while reduction in host tree growth may negatively influence mycorrhizal fungi. In this article, we analyze a long-term data set of fungal productivity, compiled over 15 years of weekly sampling during the fruiting season of a Pinus pinaster Ait. forest growing under a continental Mediterranean climate in central Spain. This data set allows us to formulate several basic questions about the role of climate in influencing epigeous fungal productivity under Mediterranean climatic conditions: (i) What is the relationship between climate and fungal yields? Our hypothesis is that fungi will respond to water availability, but also to temperature, particularly at the end of the fruiting season, and also to the AED, particularly in the warm season (ii) Does climate affect saprotrophic and mycorrhizal fungi differently? We hypothesize that their responses will differ due to the ecological constraints of their different feeding strategies. (iii) Do changes in fruiting time phenology affect fungal production or does phenology only reflect environmental constraints? and (iv) How might projected climate scenarios affect future mushroom yields?
Materials and methods
Study area This research was conducted in the south of the province of Soria, Castilla y Le on region, in central Spain. Altitude ranges from 1000 m to 1200 m a.s.l., and climate is subhumid Mediterranean but affected by continental features. Records of mean monthly temperatures and total monthly precipitation were obtained from the Soria meteorological station (41°460 N, 02°280 W; 1082 m altitude, 40 km away) for the period 1948– 2011. The data were carefully quality-controlled and homogenized using relative methods by means of neighboring stations (Vicente-Serrano et al., 2014c). Annual mean temperature is 10.4 °C, the coldest month being January (mean daily minimum temperature of 1.8 °C) and the warmest July (mean daily maximum temperature of 28.1 °C). Average annual rainfall is 556 mm, with a summer drought period typically occurring from mid-July to August (Fig. 1). Soils are arenosols and regosols developed over Tertiary and Quaternary sands, which are characterized by an excessive permeability and low nutrient content. Vegetation is dominated by Pinus pinaster Ait. Understorey is formed by different shrubs (Cistus laurifolus L., Juniperus communis L., Erica arborea L., Calluna vulgaris (L.) Hull.) and Quercus pyrenaica Willd. resprouts. The rotation of P. pinaster is 80 years, with trees cut in two phases. Until the 1970s, resin was the main product of these forests, but the decline of its market led to a shift toward wood production. Recently, mushroom harvesting has become a major activity, focused primarily on Lactarius deliciosus (L.) Gray, but also including Hygrophorus (Fr.) and Tricholoma (Fr.) Staude species. © 2015 John Wiley & Sons Ltd, Global Change Biology, 21, 3499–3510
D R Y N E S S D R I V E N D E C L I N E I N M E D I T E R R A N E A N F U N G I 3501 (a)
appropriate keys and monographs including Dennis (1968), Moser (1983), Breitenbach & Kr€ anzlin (1984, 1986, 1991, 1995, 2000, 2005), and J€ ulich (1989). The collected specimens were dehydrated and deposited at the JCyL-FUNGI herbarium in the CIF Valonsadero (Soria, Spain) available in GBIF (http:// data.gbif.org/datasets/resource/7925). Species were attributed to trophic groups based on genus level criteria according to existing literature (Agerer, 2006; Rinaldi et al., 2008; Tedersoo et al., 2010).
Climate change models
(b)
Fig. 1 (a) Mean monthly precipitation and ETo in the observatory of Soria. Dark redline: average ETo, blue line: average precipitation, light blue surface: water surplus, orange surface: water deficit. Vertical lines represent one standard deviation for average precipitation and ETo. (b) Aspect of one of the 21 sampling plots.
Sampling design To avoid tree stand age effects, we performed a random stratified sampling design. On a 1 : 20 000 scale map, we superimposed a 1-ha grid, classifying every cell into one of seven age classes, according to the forest management plan: 0–10 (regenerating but with a proportion of parent trees), 11–20, 21–40, 41–60, 61–90, and older than 90 years old. Every grid cell was numbered, and cells corresponding with every age class were selected randomly; three plots per age class were chosen, providing a total of 21 sampling plots. Selected cells were located in the field and were checked to confirm whether they met the strata assumptions, and if they did not, another random cell was selected. Within each selected cell, a sampling plot of 150 m2 (5 9 30 m) was created. Plots were fenced to prevent harvesting and trampling. Sampling was performed from September to December (week 35–50) on a weekly basis from 1997 to 2011, since this period corresponds with most of the sporocarps’ emergence. Epigeous sporocarps were collected, fresh-weighed, and identified to species level using morphological features with
© 2015 John Wiley & Sons Ltd, Global Change Biology, 21, 3499–3510
To determine the possible influence of projected climate change processes in mushroom yields, we used climate simulations from nine high resolution regional climate models (RCMs) driven by different general circulation models (GCMs), which are forced with the intermediate-emission A1B scenario (IPCC, 2013). The models were chosen based on the availability of simulations under the A1B emission scenario for the 2020–2080 period and were undertaken within the framework of the European coordinated project ENSEMBLES (http://ensemblesrt3.dmi.dk/). The names and driving GCMs of the nine RCMs used are indicated in Table S1. Models have a resolution of approximately 25 km in the study domain and use 1961–1990 as the control period. Combining results from different RCMs is advantageous for projecting climate change on the regional scale as it allows determining the most accurate models to reproduce the climate conditions in the region of interest and to select the most suitable models for projections of each meteorological variable, reducing the uncertainty of projections (Meehl & Tebaldi, 2004; Haylock et al., 2006). We used different metrics to assess the goodness of the models to reproduce observed data. For each model, we computed the monthly average, standard deviation and 10th and 90th percentiles for precipitation and maximum and minimum temperature, for the control period (1961–1990). For each model, we then calculated the average differences between these parameters and the actual values obtained from observations from Soria over the same period. We calculated the mean absolute difference and the bias between models and observations and selected the models which agreed most closely with recorded observations. For precipitation, the selected models were HC, KNMI, and MPI; for maximum temperature, they were C4I, ETHZ, and HC; and for minimum temperature, they were C4I, ETHZ, and SMHI. We created average precipitation and maximum and minimum temperature series for the selected models and calculated the average monthly difference (in % for precipitation and in °C for temperature) with the series from the observatory in Soria, obtaining coefficients that could be used to correct the multimodel series for the control period (1961–1990) and the future scenarios (2020–2080). Following this approach, we obtained projected series that matched the expected magnitude and variance of precipitation and temperature in the observatory in Soria, and these allowed us to establish robust comparisons between the control period and the projected scenarios.
R E D A et al. 3502 T . AG
Statistical analysis Pearson’s correlations were conducted to identify the climatic factors that were related to mycorrhizal and saprotrophic fungal abundance and productivity series. Climatic factors considered were monthly mean temperature and precipitation, from January to November. When climatic response was significant in several consecutive months, Pearson’s correlation between fungal series and mean climatic values for the significant period was calculated. A multiple regression including significant values from Pearson’s correlation with rainfall and mean temperature was conducted to model fungal production. All possible models were compared using AIC to select the best model. When several models showed similar AICc values (DAICc < 2) (Burnham & Anderson, 2002), the most parsimonious model was selected. Statistical analyses were performed in R version 3.1.0 (R Core Team, 2013). To explore the differential effect of other temperature parameters, additional correlations were established between fungal series and minimum and maximum monthly temperature, as well as with temperature amplitude (monthly maximum minus monthly minimum temperature). We also explored the effect of rainfall (P) and water balance (P-AED) on mushroom productivity at different temporal scales (1–12 months). For this purpose, we estimated the AED by means of the Reference Evapotranspiration (ETo) Hargreaves formulation (Hargreaves & Samani, 1985). This is based on maximum and temperature data and provides accurate estimates of the AED in Spain (Vicente-Serrano et al., 2014a). Potential lagged effects in fungal abundance and productivity time series were evaluated by estimating first and second order autocorrelations for fungal series observed data, as well as for model residuals. Fungal phenology was calculated for each trophic group. Starting date was estimated as the date when 2.5% of fungal biomass has been produced, ending date was estimated as the date where 97.5% of fungal biomass had been produced, and season length was calculated as the difference between ending date and starting date. We selected this approach for phenological traits estimation because it is less sensitive to outliers and allows comparison with previous works (Kauserud et al., 2012). Pearson’s correlations were conducted to explore whether fungal abundance and productivity was related to phenology. Phenological traits were correlated to observed values and to residual time series, calculated as the difference between observed and predicted values divided by predicted values. We projected the most plausible future fungal production for each trophic guild for the 2021–2080 period, by obtaining regression models for future climatic scenarios using the projected climate series from the multimodel approach detailed above. It is expected that climate warming will increase the AED in the Mediterranean region in the future (Giorgi & Lionello, 2008; Garcıa-Ruiz et al., 2011). For this reason, although predictive models for current climate provided good results using precipitation, we also developed models considering the AED, which causes lower soil water availability to predict future scenarios under projected climatic water balance (PAED). Therefore, we built two different models to predict mushroom productivity for current climate and future scenar-
ios, including (i) rainfall and (ii) climatic water balance. We compared the annual production between the control period and between the periods 2021–2050 and 2051–2080 and calculated the expected changes in average mushroom productivity.
Results A total of 34 840 epigeous sporocarps weighing 299 kg and belonging to at least 170 species were collected during the 15-year study period. Mycorrhizal species dominated the fungal community (96 species, 76.8% of sporocarps, 90.7% of biomass). Mycorrhizal fungi productivity ranged from 0.23 to 146.22 kg ha1 year1 (57.40 46.41 kg ha1 year1; mean SD), while their abundance ranged from 48 to 14 063 sporocarps ha1 year1 (5659 4459 sporocarps ha1 year1). Saprotrophic fungi productivity and abundance were both much lower, ranging from 0.02 to 14.51 kg ha1 year1 (5.70 4.55 kg ha1 year1) and 48 to 3838 sporocarps ha1 year1 (1714 1013 sporocarps ha1 year1), respectively. Annual variability in mushroom abundance and productivity was high (Fig. 2) with coefficients of variation that ranged from 59.1% for number of saprotrophic fungi sporocarps to 79.7% for mycorrhizal fungi weight. In spite of this variability, mushroom number and productivity showed a decreasing trend during the study period that was significant in all cases except for mycorrhizal fungi productivity (Fig. 2). Abundance and productivity were highly correlated for mycorrhizal species (r2 = 0.75; P < 0.001) and moderately so far saprotrophic species (r2 = 0.43; P = 0.008), between guilds correlation was moderately significant for abundance (r2 = 0.48; P = 0.004) and only marginally significant for productivity (r2 = 0.23; P = 0.071). Fruiting body appearance was concentrated in a very short period with production and abundance showing a sharp increase between the 41st and 42nd weeks, and decreasing rapidly in the 49th week (Fig. 3a,b). Although no significant phenological differences between guilds existed, mean starting date was a week earlier for saprotrophs than for mycorrhizae (mean se; S = 40.4 0.6 w; M = 41.7 0 .5 w), while ending date was half a week later for mycorrhizae (S = 47.5 0.5 w; M = 48 0.5 w). Overall, saprotrophic fungi therefore had a slightly longer fruiting season (S = 7.1 0.7 w; M = 6.3 0.6 w). Pearson’s correlation and multiple regressions revealed strong links between climate and fungal abundance and productivity. Saprotrophic fungi abundance was positively correlated with March, September, and December Tmean and July accumulated precipitation and negatively correlated with February–March rainfall © 2015 John Wiley & Sons Ltd, Global Change Biology, 21, 3499–3510
Abundance (Sporocarps ha−1)
D R Y N E S S D R I V E N D E C L I N E I N M E D I T E R R A N E A N F U N G I 3503
Mycorrhizal
4000 3000
1500
Saprotrophic
1000
2000 500 1000 0
0 1997 1999 2001 2003 2005 2007 2009 2011
Productivity (kg ha−1)
150
1997 1999 2001 2003 2005 2007 2009 2011 15
100
10
50
5
0
Observed Rainfall model Water balance model
0 1997 1999 2001 2003 2005 2007 2009 2011
1997 1999 2001 2003 2005 2007 2009 2011
Year
Year
(a)
600
250 200 150
400
100 200
50
0
0 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53
Productivity (kg ha−1)
10
1.0
(b) 8
0.8
6
0.6
4
0.4
2
0.2
0
Saprotrophic
Mycorrhizal Saprotrophic
Abundance (Sporocarp ha−1)
300 800
Productivity (kg ha−1)
Mycorrhizal
Abundance (Sporocarp ha−1)
Fig. 2 Observed (gray lines) and fitted models (black lines) for mycorrhizal and saprotrophic mushroom production for the studied period (1997–2011). Weight models include rainfall (solid black line) and water balance (dotted black line) models. Only rainfall models are built for abundance (number) models.
0.0 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53
Week Fig. 3 Mean weekly mushroom productivity (a) and abundance (b) for mycorrhizal (dashed line) and saprotrophic (solid line) fungi for the studied period (1997–2011).
(Fig. 4a,b). The best regression model for saprotrophic abundance was highly significant (r2 = 0.70; P = 0.004) and included a positive effect of March and September Tmean and a negative effect of February–March precipitation (Table 1, Fig. 2a). Mycorrhizal fungi abundance © 2015 John Wiley & Sons Ltd, Global Change Biology, 21, 3499–3510
was positively correlated with March Tmean and with August to September precipitation (Fig. 4a,b), and the best regression model was highly significant (r2 = 0.71; P < 0.001), when just the positive effects of August to September accumulated precipitation were included
R E D A et al. 3504 T . AG 1.0
1.0
Mean T
Pearson's r
0.8 0.6
0.6
0.4
0.4
0.2
0.2
0.0
0.0
–0.2
–0.2
Number
–0.4 –0.6
P
0.8
Number
–0.4 –0.6
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 1.0
Feb Mar Apr May Jun
Jul Aug Sep Oct Nov Dec 8-9
1.0
Mean T
0.8
Pearson's r
Jan
0.6
0.4
0.4
0.2
0.2
0.0
0.0
–0.2
–0.2
–0.4
Weight
–0.6
P
0.8
0.6
2-3
–0.4
Weight
–0.6 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Jan
Feb Mar Apr May Jun
Jul Aug Sep Oct Nov Dec 7-8
7-9
Fig. 4 Pearson’s correlation between monthly mean temperature and monthly accumulated precipitation and saprotrophic (black) and mycorrhizal (white) mushroom abundance (based on sporocarp density) and productivity (based on sporocarp weight) for the studied period (1997–2011). Numbers on the x-axis indicate mean values for several months. Horizontal lines indicate significance levels at P < 0.05 (solid lines) and P < 0.01 (dashed lines).
Table 1 Best linear regression models for mycorrhizal and saprotrophic mushrooms productivity and abundance based on AIC selection criteria
Mycorrhizal productivity Intercept September P March Tmean Saprotrophic productivity Intercept March Tmean July to September P Mycorrhizal abundance Intercept August to September P Saprotrophic abundance Intercept February to March P March Tmean September Tmean
Estimate
SE
F
P
Model
23546.7 507.0 3000.9
11284.2 112.8 1515.2
2.087 4.494 1.980
0.059
Increased evapotranspiration demand in a Mediterranean climate might cause a decline in fungal yields under global warming R E D A 1 , 2 , B E A T R I Z AG UEDA2, JOSE M. OLANO2,3, SERGIO M. T E R E S A AG 4 5 V I C E N T E - S E R R A N O and M A R I N A F E R N AN D E Z - T O I R AN 2 1 Genius Medioambiente, C/.Campaneros 26, 42200 Almazan, Soria, Spain, Area de Botanica, Departamento de Ciencias Agroforestales, EU de Ingenierıas Agrarias, Universidad de Valladolid, Campus Duques de Soria, 42004 Soria, Spain, 3Sustainable Forest Management Research Institute, Universidad de Valladolid and INIA, Avda. de Madrid 44, 34004 Palencia, Spain, 4 ~ana 1005, 50059 Instituto Pirenaico de Ecologıa, Consejo Superior de Investigaciones Cientıficas (IPE-CSIC), Avda. Montan Zaragoza, Spain, 5Departamento de Produccion Vegetal y Recursos Forestales, EU de Ingenierıas Agrarias, Universidad de Valladolid, Campus Duques de Soria, 42004 Soria, Spain
Abstract Wild fungi play a critical role in forest ecosystems, and its recollection is a relevant economic activity. Understanding fungal response to climate is necessary in order to predict future fungal production in Mediterranean forests under climate change scenarios. We used a 15-year data set to model the relationship between climate and epigeous fungal abundance and productivity, for mycorrhizal and saprotrophic guilds in a Mediterranean pine forest. The obtained models were used to predict fungal productivity for the 2021–2080 period by means of regional climate change models. Simple models based on early spring temperature and summer–autumn rainfall could provide accurate estimates for fungal abundance and productivity. Models including rainfall and climatic water balance showed similar results and explanatory power for the analyzed 15-year period. However, their predictions for the 2021–2080 period diverged. Rainfall-based models predicted a maintenance of fungal yield, whereas water balance-based models predicted a steady decrease of fungal productivity under a global warming scenario. Under Mediterranean conditions fungi responded to weather conditions in two distinct periods: early spring and late summer–autumn, suggesting a bimodal pattern of growth. Saprotrophic and mycorrhizal fungi showed differences in the climatic control. Increased atmospheric evaporative demand due to global warming might lead to a drop in fungal yields during the 21st century. Keywords: bimodal pattern, climate change, global warming, long-term monitoring, mushrooms production, mycorrhizal fungi, phenology, saprotrophic fungi Received 6 March 2015 and accepted 10 April 2015
Introduction Fungi play a critical role in forest ecosystems, decomposing organic matter and turning it into inorganic components accessible to tree roots. Moreover, mycorrhizal fungi form symbiotic associations with tree roots, improving trees’ water and nutrient availability while providing defense against pathogens, thus boosting tree growth and survival rates. Beyond the provision of these basic ecosystem services, mushrooms are also a valuable source of wild food. Mushroom-picking is the basis for a large economic sector including mushroom transformation and commercialization, and even cre ates important benefits for the tourist sector (Agreda et al., 2014).
Correspondence: Jose M. Olano, tel. +34 975 129 485, fax +34 975 129 201, e-mail: [email protected]
© 2015 John Wiley & Sons Ltd
A sustainable management of wild mushrooms requires a thorough comprehension of the factors driving mushroom production and temporal variability. Control of mushroom production is related to multiple environmental factors such as site soil characteristics (Barroetave~ na et al., 2008; Alonso Ponce et al., 2014), endogenous dynamics of fungal communities (Smith et al., 2002; De Miguel et al., 2014), succession (Fern andez-Toiran et al., 2006; Baptista et al., 2010), forest management (Egli et al., 2006; Le Tacon et al., 2014), and climate (Baptista et al., 2010; Diez et al., 2013). Climate is a major component determining mushroom production, particularly during the fruiting period (Boddy et al., 2014). Increases in temperature and modification of rainfall patterns may have already altered mushroom phenology (Kauserud et al., 2008; B€ untgen et al., 2013b; Diez et al., 2013; Gange et al., 2013), as well as mushroom productivity (B€ untgen et al., 2012b), with positive and negative outcomes being detected 3499
R E D A et al. 3500 T . AG depending on the study region (B€ untgen et al., 2012a; Sato et al., 2012). Most fungi produce ephemeral fruiting bodies that can be observed only for a few days each year, making the acquisition of accurate annual productivity data time-consuming and challenging. In fact, progress in unveiling the climatic forcing of fungal productivity has been hindered because of the scarcity of long-term time series based on systematic sampling (Zambonelli et al., 2012; Boddy et al., 2014). Indirect estimates based on market prices of commercial fungi have been used to estimate abundance and relate it to climate (B€ untgen et al., 2012a; Le Tacon et al., 2014), while national inventories based on collaborative work also provide some information on fungal relative abundance and phenology (Diez et al., 2013; Boddy et al., 2014). However, the value of these approaches is often limited by inconsistency in the sampling effort or design; more accurate prediction requires high-quality time series data sets based on the systematic sampling of fungal productivity. However, existing time series mostly comprise relatively short periods (Salerni et al., 2002; Bonet et al., 2008), with longer records being scarce (Egli et al., 2006; Martınez-Pe~ na et al., 2012; Agreda et al., 2014) and missing some biogeographical areas such as the Mediterranean. Climate warming has been strong in the past five decades in the western Mediterranean region (Alpert et al., 2008; Bertin, 2008) with a range of increase between 0.2° and 0.4 °C decade1, which has dramatically increased the atmospheric evaporative demand (AED) (Vicente-Serrano et al., 2014a). Current climate change models predict an intense temperature increase for the coming decades that ranges from 2 to 3 °C for the 2040–2070 period (Giorgi, 2006; GarcıaRuiz et al., 2011) that would cause higher AED and a general decrease of water resources (Garcıa-Ruiz et al., 2011; Vicente-Serrano et al., 2014b). Although the projections for precipitation are less reliable than for temperature (Giorgi et al., 2004; Deque et al., 2005), most studies project a general tendency toward less precipitation in the next century (Giorgi & Lionello, 2008; Evans, 2009). The expected higher AED and lower precipitation mean that more frequent and severe drought events are therefore predicted (Dai, 2013; Cook et al., 2014). A decrease in water availability during the growing season would severely impact Mediterranean forest ecosystems as it would reduce tree growth and increase tree mortality (Carnicer et al., 2011; Pasho et al., 2011). Mediterranean mushrooms may also be affected by increasing aridity; directly so, because fungal growth and sporocarp production is closely linked to soil humidity (Murat et al., 2008), but also indirectly because reductions in
primary productivity will limit resource availability to saprotrophic fungi, while reduction in host tree growth may negatively influence mycorrhizal fungi. In this article, we analyze a long-term data set of fungal productivity, compiled over 15 years of weekly sampling during the fruiting season of a Pinus pinaster Ait. forest growing under a continental Mediterranean climate in central Spain. This data set allows us to formulate several basic questions about the role of climate in influencing epigeous fungal productivity under Mediterranean climatic conditions: (i) What is the relationship between climate and fungal yields? Our hypothesis is that fungi will respond to water availability, but also to temperature, particularly at the end of the fruiting season, and also to the AED, particularly in the warm season (ii) Does climate affect saprotrophic and mycorrhizal fungi differently? We hypothesize that their responses will differ due to the ecological constraints of their different feeding strategies. (iii) Do changes in fruiting time phenology affect fungal production or does phenology only reflect environmental constraints? and (iv) How might projected climate scenarios affect future mushroom yields?
Materials and methods
Study area This research was conducted in the south of the province of Soria, Castilla y Le on region, in central Spain. Altitude ranges from 1000 m to 1200 m a.s.l., and climate is subhumid Mediterranean but affected by continental features. Records of mean monthly temperatures and total monthly precipitation were obtained from the Soria meteorological station (41°460 N, 02°280 W; 1082 m altitude, 40 km away) for the period 1948– 2011. The data were carefully quality-controlled and homogenized using relative methods by means of neighboring stations (Vicente-Serrano et al., 2014c). Annual mean temperature is 10.4 °C, the coldest month being January (mean daily minimum temperature of 1.8 °C) and the warmest July (mean daily maximum temperature of 28.1 °C). Average annual rainfall is 556 mm, with a summer drought period typically occurring from mid-July to August (Fig. 1). Soils are arenosols and regosols developed over Tertiary and Quaternary sands, which are characterized by an excessive permeability and low nutrient content. Vegetation is dominated by Pinus pinaster Ait. Understorey is formed by different shrubs (Cistus laurifolus L., Juniperus communis L., Erica arborea L., Calluna vulgaris (L.) Hull.) and Quercus pyrenaica Willd. resprouts. The rotation of P. pinaster is 80 years, with trees cut in two phases. Until the 1970s, resin was the main product of these forests, but the decline of its market led to a shift toward wood production. Recently, mushroom harvesting has become a major activity, focused primarily on Lactarius deliciosus (L.) Gray, but also including Hygrophorus (Fr.) and Tricholoma (Fr.) Staude species. © 2015 John Wiley & Sons Ltd, Global Change Biology, 21, 3499–3510
D R Y N E S S D R I V E N D E C L I N E I N M E D I T E R R A N E A N F U N G I 3501 (a)
appropriate keys and monographs including Dennis (1968), Moser (1983), Breitenbach & Kr€ anzlin (1984, 1986, 1991, 1995, 2000, 2005), and J€ ulich (1989). The collected specimens were dehydrated and deposited at the JCyL-FUNGI herbarium in the CIF Valonsadero (Soria, Spain) available in GBIF (http:// data.gbif.org/datasets/resource/7925). Species were attributed to trophic groups based on genus level criteria according to existing literature (Agerer, 2006; Rinaldi et al., 2008; Tedersoo et al., 2010).
Climate change models
(b)
Fig. 1 (a) Mean monthly precipitation and ETo in the observatory of Soria. Dark redline: average ETo, blue line: average precipitation, light blue surface: water surplus, orange surface: water deficit. Vertical lines represent one standard deviation for average precipitation and ETo. (b) Aspect of one of the 21 sampling plots.
Sampling design To avoid tree stand age effects, we performed a random stratified sampling design. On a 1 : 20 000 scale map, we superimposed a 1-ha grid, classifying every cell into one of seven age classes, according to the forest management plan: 0–10 (regenerating but with a proportion of parent trees), 11–20, 21–40, 41–60, 61–90, and older than 90 years old. Every grid cell was numbered, and cells corresponding with every age class were selected randomly; three plots per age class were chosen, providing a total of 21 sampling plots. Selected cells were located in the field and were checked to confirm whether they met the strata assumptions, and if they did not, another random cell was selected. Within each selected cell, a sampling plot of 150 m2 (5 9 30 m) was created. Plots were fenced to prevent harvesting and trampling. Sampling was performed from September to December (week 35–50) on a weekly basis from 1997 to 2011, since this period corresponds with most of the sporocarps’ emergence. Epigeous sporocarps were collected, fresh-weighed, and identified to species level using morphological features with
© 2015 John Wiley & Sons Ltd, Global Change Biology, 21, 3499–3510
To determine the possible influence of projected climate change processes in mushroom yields, we used climate simulations from nine high resolution regional climate models (RCMs) driven by different general circulation models (GCMs), which are forced with the intermediate-emission A1B scenario (IPCC, 2013). The models were chosen based on the availability of simulations under the A1B emission scenario for the 2020–2080 period and were undertaken within the framework of the European coordinated project ENSEMBLES (http://ensemblesrt3.dmi.dk/). The names and driving GCMs of the nine RCMs used are indicated in Table S1. Models have a resolution of approximately 25 km in the study domain and use 1961–1990 as the control period. Combining results from different RCMs is advantageous for projecting climate change on the regional scale as it allows determining the most accurate models to reproduce the climate conditions in the region of interest and to select the most suitable models for projections of each meteorological variable, reducing the uncertainty of projections (Meehl & Tebaldi, 2004; Haylock et al., 2006). We used different metrics to assess the goodness of the models to reproduce observed data. For each model, we computed the monthly average, standard deviation and 10th and 90th percentiles for precipitation and maximum and minimum temperature, for the control period (1961–1990). For each model, we then calculated the average differences between these parameters and the actual values obtained from observations from Soria over the same period. We calculated the mean absolute difference and the bias between models and observations and selected the models which agreed most closely with recorded observations. For precipitation, the selected models were HC, KNMI, and MPI; for maximum temperature, they were C4I, ETHZ, and HC; and for minimum temperature, they were C4I, ETHZ, and SMHI. We created average precipitation and maximum and minimum temperature series for the selected models and calculated the average monthly difference (in % for precipitation and in °C for temperature) with the series from the observatory in Soria, obtaining coefficients that could be used to correct the multimodel series for the control period (1961–1990) and the future scenarios (2020–2080). Following this approach, we obtained projected series that matched the expected magnitude and variance of precipitation and temperature in the observatory in Soria, and these allowed us to establish robust comparisons between the control period and the projected scenarios.
R E D A et al. 3502 T . AG
Statistical analysis Pearson’s correlations were conducted to identify the climatic factors that were related to mycorrhizal and saprotrophic fungal abundance and productivity series. Climatic factors considered were monthly mean temperature and precipitation, from January to November. When climatic response was significant in several consecutive months, Pearson’s correlation between fungal series and mean climatic values for the significant period was calculated. A multiple regression including significant values from Pearson’s correlation with rainfall and mean temperature was conducted to model fungal production. All possible models were compared using AIC to select the best model. When several models showed similar AICc values (DAICc < 2) (Burnham & Anderson, 2002), the most parsimonious model was selected. Statistical analyses were performed in R version 3.1.0 (R Core Team, 2013). To explore the differential effect of other temperature parameters, additional correlations were established between fungal series and minimum and maximum monthly temperature, as well as with temperature amplitude (monthly maximum minus monthly minimum temperature). We also explored the effect of rainfall (P) and water balance (P-AED) on mushroom productivity at different temporal scales (1–12 months). For this purpose, we estimated the AED by means of the Reference Evapotranspiration (ETo) Hargreaves formulation (Hargreaves & Samani, 1985). This is based on maximum and temperature data and provides accurate estimates of the AED in Spain (Vicente-Serrano et al., 2014a). Potential lagged effects in fungal abundance and productivity time series were evaluated by estimating first and second order autocorrelations for fungal series observed data, as well as for model residuals. Fungal phenology was calculated for each trophic group. Starting date was estimated as the date when 2.5% of fungal biomass has been produced, ending date was estimated as the date where 97.5% of fungal biomass had been produced, and season length was calculated as the difference between ending date and starting date. We selected this approach for phenological traits estimation because it is less sensitive to outliers and allows comparison with previous works (Kauserud et al., 2012). Pearson’s correlations were conducted to explore whether fungal abundance and productivity was related to phenology. Phenological traits were correlated to observed values and to residual time series, calculated as the difference between observed and predicted values divided by predicted values. We projected the most plausible future fungal production for each trophic guild for the 2021–2080 period, by obtaining regression models for future climatic scenarios using the projected climate series from the multimodel approach detailed above. It is expected that climate warming will increase the AED in the Mediterranean region in the future (Giorgi & Lionello, 2008; Garcıa-Ruiz et al., 2011). For this reason, although predictive models for current climate provided good results using precipitation, we also developed models considering the AED, which causes lower soil water availability to predict future scenarios under projected climatic water balance (PAED). Therefore, we built two different models to predict mushroom productivity for current climate and future scenar-
ios, including (i) rainfall and (ii) climatic water balance. We compared the annual production between the control period and between the periods 2021–2050 and 2051–2080 and calculated the expected changes in average mushroom productivity.
Results A total of 34 840 epigeous sporocarps weighing 299 kg and belonging to at least 170 species were collected during the 15-year study period. Mycorrhizal species dominated the fungal community (96 species, 76.8% of sporocarps, 90.7% of biomass). Mycorrhizal fungi productivity ranged from 0.23 to 146.22 kg ha1 year1 (57.40 46.41 kg ha1 year1; mean SD), while their abundance ranged from 48 to 14 063 sporocarps ha1 year1 (5659 4459 sporocarps ha1 year1). Saprotrophic fungi productivity and abundance were both much lower, ranging from 0.02 to 14.51 kg ha1 year1 (5.70 4.55 kg ha1 year1) and 48 to 3838 sporocarps ha1 year1 (1714 1013 sporocarps ha1 year1), respectively. Annual variability in mushroom abundance and productivity was high (Fig. 2) with coefficients of variation that ranged from 59.1% for number of saprotrophic fungi sporocarps to 79.7% for mycorrhizal fungi weight. In spite of this variability, mushroom number and productivity showed a decreasing trend during the study period that was significant in all cases except for mycorrhizal fungi productivity (Fig. 2). Abundance and productivity were highly correlated for mycorrhizal species (r2 = 0.75; P < 0.001) and moderately so far saprotrophic species (r2 = 0.43; P = 0.008), between guilds correlation was moderately significant for abundance (r2 = 0.48; P = 0.004) and only marginally significant for productivity (r2 = 0.23; P = 0.071). Fruiting body appearance was concentrated in a very short period with production and abundance showing a sharp increase between the 41st and 42nd weeks, and decreasing rapidly in the 49th week (Fig. 3a,b). Although no significant phenological differences between guilds existed, mean starting date was a week earlier for saprotrophs than for mycorrhizae (mean se; S = 40.4 0.6 w; M = 41.7 0 .5 w), while ending date was half a week later for mycorrhizae (S = 47.5 0.5 w; M = 48 0.5 w). Overall, saprotrophic fungi therefore had a slightly longer fruiting season (S = 7.1 0.7 w; M = 6.3 0.6 w). Pearson’s correlation and multiple regressions revealed strong links between climate and fungal abundance and productivity. Saprotrophic fungi abundance was positively correlated with March, September, and December Tmean and July accumulated precipitation and negatively correlated with February–March rainfall © 2015 John Wiley & Sons Ltd, Global Change Biology, 21, 3499–3510
Abundance (Sporocarps ha−1)
D R Y N E S S D R I V E N D E C L I N E I N M E D I T E R R A N E A N F U N G I 3503
Mycorrhizal
4000 3000
1500
Saprotrophic
1000
2000 500 1000 0
0 1997 1999 2001 2003 2005 2007 2009 2011
Productivity (kg ha−1)
150
1997 1999 2001 2003 2005 2007 2009 2011 15
100
10
50
5
0
Observed Rainfall model Water balance model
0 1997 1999 2001 2003 2005 2007 2009 2011
1997 1999 2001 2003 2005 2007 2009 2011
Year
Year
(a)
600
250 200 150
400
100 200
50
0
0 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53
Productivity (kg ha−1)
10
1.0
(b) 8
0.8
6
0.6
4
0.4
2
0.2
0
Saprotrophic
Mycorrhizal Saprotrophic
Abundance (Sporocarp ha−1)
300 800
Productivity (kg ha−1)
Mycorrhizal
Abundance (Sporocarp ha−1)
Fig. 2 Observed (gray lines) and fitted models (black lines) for mycorrhizal and saprotrophic mushroom production for the studied period (1997–2011). Weight models include rainfall (solid black line) and water balance (dotted black line) models. Only rainfall models are built for abundance (number) models.
0.0 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53
Week Fig. 3 Mean weekly mushroom productivity (a) and abundance (b) for mycorrhizal (dashed line) and saprotrophic (solid line) fungi for the studied period (1997–2011).
(Fig. 4a,b). The best regression model for saprotrophic abundance was highly significant (r2 = 0.70; P = 0.004) and included a positive effect of March and September Tmean and a negative effect of February–March precipitation (Table 1, Fig. 2a). Mycorrhizal fungi abundance © 2015 John Wiley & Sons Ltd, Global Change Biology, 21, 3499–3510
was positively correlated with March Tmean and with August to September precipitation (Fig. 4a,b), and the best regression model was highly significant (r2 = 0.71; P < 0.001), when just the positive effects of August to September accumulated precipitation were included
R E D A et al. 3504 T . AG 1.0
1.0
Mean T
Pearson's r
0.8 0.6
0.6
0.4
0.4
0.2
0.2
0.0
0.0
–0.2
–0.2
Number
–0.4 –0.6
P
0.8
Number
–0.4 –0.6
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 1.0
Feb Mar Apr May Jun
Jul Aug Sep Oct Nov Dec 8-9
1.0
Mean T
0.8
Pearson's r
Jan
0.6
0.4
0.4
0.2
0.2
0.0
0.0
–0.2
–0.2
–0.4
Weight
–0.6
P
0.8
0.6
2-3
–0.4
Weight
–0.6 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Jan
Feb Mar Apr May Jun
Jul Aug Sep Oct Nov Dec 7-8
7-9
Fig. 4 Pearson’s correlation between monthly mean temperature and monthly accumulated precipitation and saprotrophic (black) and mycorrhizal (white) mushroom abundance (based on sporocarp density) and productivity (based on sporocarp weight) for the studied period (1997–2011). Numbers on the x-axis indicate mean values for several months. Horizontal lines indicate significance levels at P < 0.05 (solid lines) and P < 0.01 (dashed lines).
Table 1 Best linear regression models for mycorrhizal and saprotrophic mushrooms productivity and abundance based on AIC selection criteria
Mycorrhizal productivity Intercept September P March Tmean Saprotrophic productivity Intercept March Tmean July to September P Mycorrhizal abundance Intercept August to September P Saprotrophic abundance Intercept February to March P March Tmean September Tmean
Estimate
SE
F
P
Model
23546.7 507.0 3000.9
11284.2 112.8 1515.2
2.087 4.494 1.980
0.059
