Arbuscular mycorrhizal fungi persist in dying Euphorbia ingens trees

a Department of Genetics, Forestry and Agricultural Biotechnology Institute (FABI), Faculty of Natural and Agricultural Sciences (NAS), University of Pretoria, Pretoria, South Africa b South African Environmental Observation Network, Arid Lands Node, Kimberley, South Africa c Department of Biochemistry and Microbiology, Mycorrhizal Research Laboratory, Rhodes University, Grahamstown, South Africa d Department of Microbiology, FABI, NAS, University of Pretoria, South Africa e Department of Plant and Soil Sciences, FABI, NAS, University of Pretoria, Pretoria, South Africa


Introduction
Large-scale forest declines have been reported the last decade, and are expected to increase due to the ongoing environmental changes (Allen et al., 2010;Anderegg et al., 2013). Climatic extremes trigger tree die-back due to physiological stresses, yet individuals are most likely finally killed-off by coinciding pest and pathogen outbreaks (Desprez-Loustau et al., 2006;McDowell et al., 2011;Anderegg et al., 2015). During and south side of each tree. The soil samples were collected from the top 20 cm of the rhizospheric soil with a soil core of 7 cm in diameter. The soil samples were air-dried and stored at 4 ºC until they were analysed.

AM fungal assessment 2.3.1. AM fungal colonisation of the roots
To assess AM colonisation, root samples were washed, cleared and stained according to a modified method of Koske and Gemma (1989). Roots were rinsed and cut into 1-3 cm fragments. The fragments were cleared in 5 % KOH at 90 ºC for 30 min, then bleached in alkaline H 2 O 2 for 10 min. The roots were acidified with 0.1 M HCl for 2 h and stained in lactoglycerol (lactic acid, glycerol, water 13:12:16 (v/v/v)) containing 0.05 % trypan blue, at 90 ºC for 30 min. Finally, the roots were destained in lactoglycerol for 12 h. Twenty root fragments were randomly selected from each tree, these were mounted on a microscope slide and examined using a light microscope (Zeiss Axioskop 2 Plus, Oberkochen, Germany).
The percentage of AM colonisation was estimated according to Trouvelot (1986), and the following parameters were recorded: frequency of AM in the root system (F%); intensity of AM colonisation in the root system (M%); intensity of AM colonisation in root fragments (m%). These parameters were calculated with Mycocalc, a free mycorrhiza measuring programme (https://www2.dijon.inra.fr/mychintec/Mycocalcprg/download.html).
To measure AM spore numbers, spores from the soil samples were extracted using a wet sieving and decanting method followed by sucrose centrifugation (Schenck, 1982;Smith and Dickson, 1997). To extract the spores, soil samples were sieved through a 2 mm mesh. The sieved soil (100 g) was stirred with 200 ml of water for 5 min and settled for 15 sec. The supernatant obtained was decanted through a nest of soil sieves (425 µm, 250 µm, 125 µm, 50 µm), and the remaining debris per soil sieve was collected into centrifuge tubes with water. To purify the spores, the aqueous suspension was centrifuged (1900 g for 5 min) and the supernatant discarded. The debris was resuspended in 60 % sucrose and centrifuged for 5 min (1900 g). The supernatant obtained was decanted onto the 50 µm sieve and rinsed. The spores decanted onto the sieve were transferred to a filter paper using a Buchner funnel. The number of total spores was counted per sample, with spore abundance expressed as the number of spores per 100 g dry soil.

Soil properties
Soil phosphorus (P), mineral nitrogen and pH were analysed from a composite soil sample from the north and south hemisphere of each tree. Soil P was extracted from the soil samples according to the P-bray method (Bray and Kurtz, 1945) and determined by automatic colorimetric analysis. Mineral nitrogen defined as ammonium (NH 4 + ) and nitrate (NO 3 -) were extracted with 1 M KCl (SSSA, 1977) and determined by the Kjeldahl method, using Devarda alloy to reduce NO 3to NH 4 + (Keeney & Nelson, 1982). Soil pH was determined after dilution at a ratio of 1:2.5 soil:water (v/v) using a digital pH meter. The soil texture at Last Post and Enzelsberg were previously classified as sandy loam and were very rocky, whereas the soils at Capricorn are loamy sand and

Data analyses
The influence of tree health status (grey discoloration), the site and their interaction on the AM colonisation (F%, M%, m%) of the roots were analysed using linear models.
The number of AM spores was similarly analysed. To account for micro-environmental variations within the tree, the models also included the co-variation with root hemisphere (north-or south-side). Variables indicating AM colonisation were square root transformed prior to analysis to conform to normality. Model validity was also tested by visual examination of residual plots and by assessment of dispersion parameters. When an explanatory variable was significant, individual means were compared by Fisher´s least significant difference (LSD) test. The 'agricolae' package of the R software (R Core Team, 2014) was used for linear models.
To examine if soil texture (sandy loam or loamy sand), pH, NH 4 + , NO 3 -, and/or P was significantly associated with observed AM colonisation (F%, M%, m%) of the roots and AM spores, a redundancy analysis (RDA) was used. The significance of the overall ordination (test on all axes) was tested using 9999 permutations. A forward selection of variables was used to rank the most important soil properties associated with the response variables (Šmilauer and Lepš, 2014). All response variables were centred and standardized due to their varying measurement units. These statistics were calculated in CANOCO 5 (Ter Braak and Šmilauer, 2012).

AM fungal assessment
AM fungi were strongly associated with E. ingens trees, since 46 out of 48 sampled trees were colonised (Fig. 1). AM root colonisation and spore numbers significantly differed between the three sites (Table 1). Specifically, the frequency of AM fungi in the root system (F%) was significantly higher in Last Post when compared to Capricorn and Enzelsberg (Fig. 2a). AM spores were lowest in the more sandy Capricorn soils ( Fig. 2c). Interestingly, the change in tree health status only affected the spore numbers (Table 1). Categories indicating varying levels of grey discoloration only affected the number of spores in the surrounding soil ( Fig. 2d), and there was no difference in the frequency of AM fungi in the root system (F%; Fig. 2b).

Soil properties
Soil properties differed between the study sites ( Table 2). The sites at Last Post and Enzelsberg had sandy loam soils, which were both markedly rockier than the loamy sand soils found at Capricorn. The Capricorn site also had the lowest P and NH 4 + content. The site at Last Post had the lowest amount of NO 3 -. The pH of soils varied less among sites (ranged between 6.04 and 6.69), with Capricorn having the least acidic soils on average.

Relationship between soil properties and AM fungal colonisation of the roots and AM spores
Only the level of soil NO 3and soil texture significantly influenced AM colonisation (%F, %M, %m) of the roots and the number of AM spores recorded (Fig. 3). However, AM colonisation of the roots and AM spores appeared to respond along distinct soil gradients. Higher frequencies of AM fungal colonization of the roots were more directly 10 3. Results     related to soils with lower NO 3concentration. Soil texture appears to be more directly correlated with AM spore numbers. In particular, sandy loam soils were significantly associated with higher numbers of AM spores. ingens should have distinct C-dynamics compared to 'normal' C 3 savanna trees.

Discussion
Compared to C 3 photosynthesizing trees, E. ingens utilizes a markedly different photosynthetic pathway, Crassulacean acid metabolism (CAM). CAM allows plants to fix CO 2 at night when it is most available, and thus also limiting excessive water loss (Pearcy et al., 1987;Keeley and Rundel, 2003). Apart from being water-use efficient, another advantage of plants using CAM is the ability to maintain carbohydrate stocks to maintain functioning when being water stressed (Dodd et al., 2002;Winter and Holtum, 2015). The resource efficiency of E. ingens individuals may lead to more efficient utilization of carbon stocks during distressed periods, maintaining some payment of the 'cost' of preserving AM fungal benefits. Indeed, after disturbances such as branch dieback, some plants shift carbon allocation to be prioritized to maintain reserve stocks (non-structural carbohydrates), instead of plant growth (Wiley et al., 2016). Given that CAM plants are already photosynthetically efficient in drylands, the complementary role of non-structural carbon reserves being allocated to maintain crucial symbioses after environmental shocks needs further exploration.

Relationship between soil properties and AM fungal colonisation of the roots and AM spores
Site characteristics influenced AM ecology. However, AM root colonisation appeared to respond along a different soil gradient than AM spores. The frequency of AM colonisation is more likely determined by soil chemistry while the number of AM spores is related to soil texture. In particular, the frequency of AM colonisation in the root system was negatively correlated with the NO 3 -. In general, a higher N availability in the soil negatively affects fine root quality (Pregitzer et al., 1995), which could decrease the habitat for AM fungi (Treseder and Allen, 2000). Our findings support reports that AM colonisation should be higher in N-limited environments (Blanke et al., 2005). However, to our knowledge there is no defined N threshold inhibiting AM colonisation. Some studies argue that regulation of AM colonisation based on soil N involves complex mechanisms, making it difficult to generalize any observed trends (Treseder and Allen, 2002;Johnson et al., 2003;Staddon et al., 2004;Treseder, 2004).
One way forward is to determine which type of N, e.g., NH 4 + or NO 3 -, would be more likely to influence AM fungal presence and abundance.
The lowest number of AM spores was predominantly associated with sandy soils. Sandy textures favours water infiltration and low nutrient retention (Lehmann and Schroth, 2003). Thus, sandy soils might have favoured lower spore retention in comparison with the more loamy soils. Furthermore, the AM spore abundances observed in our study was considerably higher in comparison to a study conducted in the same region of our Enzelsberg study site (averaging only 80 spores/100 g of soil, Straker et al., 2007).
However, it should be noted that the lower AM spore abundance in the latter study is most likely determined by the study system, since it was done in slime dams of gold mines, where soil degradation is arguably much higher than the livestock farming at our sites. Degradation is thus another element that might affect AM spore numbers.
Nonetheless, there is increasing evidence that, in general, AM fungal communities primarily respond to prevailing habitat conditions, in particular soil heterogeneity, than to land-use conditions per se (de Carvalho et al., 2012;Hazard et al., 2013;Jansa et al., 2014;Cheeke et al., 2015). Of course, in some cases the intensity of disturbance would also influence AM fungal presence (Oehl et al., 2010).

Conclusions
Euphorbia ingens trees are symbiotically associated with AM fungi, but unexpectedly AM abundance was not influenced by levels of tree die-back. E. ingens maintains carbon allocation to roots and AM fungal structures, and in turn, the persistence of AM fungi would remain to supply nutrient benefits. This persistence could slow down tree die-back, but clearly does not prevent die-off, as perpetual loss of branches would inevitably decrease carbohydrate supply beyond critical levels (Galiano et al., 2011).
Eventually, both tree and fungus will die. However, persistence of AM fungi suggests that carbon reserves could aid host-recovery should the landscape disturbance fall short of killing the entire E. ingens population (Wiley et al., 2016). We further highlight why all efforts should be directed to mitigate large-scale declines of E. ingens populations (van der Linde et al., 2017).
The fact that there is a lack of association between AM root colonisation and the number of spores found in the adjacent soil matrix is not unexpected (López-Sánchez and Honrubia, 1992;Uhlmann et al., 2006). The number of spores obtained by sieving methods detects only those species which produce spores and not all spores observed are necessarily viable. Thus, it is conceivable that with our study design the number of AM spores could be an artefact of soil texture. This finding is important for future sampling designs and biological inference. Seasonal samplings may also provide a clearer understanding of the symbiosis. Nonetheless, for these tree-fungal interactions, local site effects most likely determined AM presence and abundance, in particular NO 3for the frequency of AM colonisation, and more loamy soils for AM spores.
AM colonisation in grasses decreased in overgrazed xeric savannas in Namibia (Uhlmann et al., 2006). Necessary future studies of E. ingens die-offs should also incorporate land-use intensity and plant host combinations to assess AM fungal ecology. Also, increased nitrogen deposition due to agricultural activities may decrease AM populations. This is worrying in times of dramatic CO 2 increases, as these fungi are crucial for carbon sequestration (Treseder and Allen, 2000). Finally, future studies based on molecular identification techniques will help to understand whether AM fungal species richness and diversity are indeed declining with E. ingens.