Fisseha Asmelash · Tamrat Bekele · Fassil Kebede ·Zerihun Belay

Abstract We investigated the arbuscular mycorrhizal fungi(AMF) status of ten nurseries suitable for restoration of dry evergreen Afromontane forests in Ethiopia. We quantif ied AMF root colonization (RC) and spore abundance (SA) in seedlings of nine native tree species namely Acacia abyssinica Hochst. ex Benth., Cordia africana Lam., Dovyalis abyssinica (A. Rich.) Warb., Hagenia abyssinica J.F. Gmel.,Juniperus procera Hochst. ex Endl., Millettia ferruginea(Hochst.) Baker, Olea europaea L. subsp. cuspidata (Wall.ex G. Don) Cif., Podocarpus falcatus (Thunb.) R. Br. ex Mirb. and Prunus africana (Hook. f.) Kalkman. We used the ink and vinegar method to stain AMF in roots. RC levels ranged from 8.00 to 99.67% and were generally higher than the RC levels reported from other similar nurseries in Ethiopia. SA levels ranged from 1 to 25 spores g −1 and were comparable with some reports from the f ield in Ethiopia but they were lower than levels reported by another similar study. RC was more aff ected by host species than nursery location, while the reverse was true for SA. The results also showed that nursery management could improve AMF status among seedlings. When all nursery tree species were considered, RC and SA levels were unrelated. No strong correlation existed between the nursery management variables considered and RC or SA. However, considering C. africana,J. procera and P. falcatus separately, RC-age ( r s = 0.829,P = 0.042) correlation for O. europaea and RC-pot diameter ( r s = 0.820, P = 0.046), RC-pot volume ( r s = 0.928,P = 0.008) and SA-age ( r s = 0.943, P = 0.005) correlations for C. africana, were signif icant, strong and positive. Generally, most of the tree species and particularly, early-mid successional tree species had suffi cient AMF inoculum. Hence,only the mid-late successional tree species; J. procera, P.falcatus, and P. africana may require AMF inoculation,preferably, during f iled planting. Based on our results, age and pot volume were identif ied to be important variables potentially aff ecting RC and SA. To better understand the Effects of these and other nursery management variables,additional study is required. We demonstrated for the f irst time that black Hero ink is suitable for staining root AMF and can be used in future AMF research.

Keywords Arbuscular mycorrhizal fungi · Dovyalis abyssinica · Dry evergreen afromontane forests · Forest restoration · Ink and vinegar staining · Tree nursery

Introduction

Ethiopia has by far the largest landmass suitable for montane forests in Africa (Bussam 2006). Most of the Ethiopian highlands were once covered by dry evergreen Afromontane forests (DAF) (Friis et al. 2010). Due to anthropogenic land-use change, however, only fragments of degraded DAF now remain (Teketay and Anders 1995;Kindu et al. 2013; Tolessa et al. 2017). Restoration of the DAF is currently the most important environmental agenda in the Ethiopian highlands.

Success in restoration is a function of the depth of local knowledge (Diggelen et al. 2001). In our case, knowledge of the plant and arbuscular mycorrhizal fungi (AMF) ecology of the DAF has widely been reported (Lemenih and Teketay 2004; Wubet et al. 2003a, 2003b, 2006, 2009;Birhane et al. 2018). The biology and restoration mechanisms of the most important DAF native trees have been documented by Negash ( 2010). The facilitative role of nurse plants for restoration of DAF native trees has been demonstrated (Aerts et al. 2007; Negash and Kagnew 2013). Despite these eff orts, little is known of the AMF status of nurseries suitable for DAF restoration. To date,the only study of such nurseries was reported by Michelsen( 1992 ), who studied four DAF nurseries in central Ethiopia. Our goal was to expand this knowledge base by studying the AMF status of seedlings in ten DAF nurseries located in central and northern Ethiopia.

Arbuscular mycorrhizal fungi of the phylum Glomeromycota (Schüßler et al. 2001) are multiple nucleated, asexually reproducing eukaryotes forming an obligate symbiotic relationship with most terrestrial plants (Wang and Qiu 2006; Schüßler et al. 2007). These fungi are key to the restoration of forests (Asmelash et al. 2016; Neuenkamp et al. 2018). However, whether AMF inoculation is important in forest restoration projects is still a matter of debate(Ryan and Graham 2018; Rillig et al. 2019). AMF are found in every soil (Abbott and Robson 1991; Brundrett and Abbott 2002; Smith and Read 2008) and a very small amount of infective AMF propagules could be enough to cause suffi cient root infection (Daft and Nicolson 1969).According to Daft and Nicolson ( 1969), inoculating tomato (Solanum lycopersicumL.) with AMF inoculum containing 3 versus 225 spores, resulted in similar AMF root infection. In contrast, Carling et al. ( 1979) and Douds et al. ( 2010) reported that increasing the amount of infective inoculum, up to a threshold level, increased AMF root infection in soybean (Glycine max(L.) Merr.) and bahiagrass (Paspalum notatumFlüggé). Urgiles et al. ( 2009)reported that some nursery seedlings grown on non-sterile substrate were not infected by AMF and showed lower growth rates compared to inoculated seedlings, which also showed higher AMF root colonization. Furthermore,inoculation success, in general, depends on the AMF status of the planting sites and planting materials (Verbruggen et al. 2013). Our study objectives were to quantify the AMF status of seedlings raised in DAF nurseries and to provide a database for use by AMF researchers and AMF inoculation eff orts in DAF restoration projects.

Materials and methods

Seedling collection

Severe loss of DAF has occurred in central and northern Ethiopia where large restoration projects have been carried out. We studied nurseries at Addis Ababa, north Shewa and south Wollo of the Amahara region, at west Shewa and west Hararge of the Oromiya region, and at south Tigray (Fig. 1).Data and seedlings were collected from May 1−30, 2018.Collecting seedlings at this time of the year enabled quantif ication of RC and SA of seedlings just before they were transplanted to f ield sites at the beginning of the rainy season in June−July.

We identif ied nurseries for native trees and shrubs in collaboration with representatives of our selected districts.Ten of these nurseries were randomly selected for this study.From the selected nurseries, 16 seedlings were collected for each evaluated species:Acacia abyssinicaHochst. ex Benth.(Fabaceae),Cordia africanaLam. (Boraginaceae),Dovyalis abyssinica(A. Rich.) Warb. (Flacourtiaceae),Hagenia abyssinicaJ.F. Gmel. (Rosaceae), Juniperus proceraHochst. ex Endl. (Cupressaceae),Millettia ferruginea(Hochst.) Baker(Fabaceae),Olea europaeaL. subsp.cuspidata(Wall. ex G.Don) Cif. (Oleaceae),Podocarpus falcatus(Thunb.) R. Br.ex Mirb. (Podocarpaceae) andPrunus africana(Hook. f.)Kalkman (Rosaceae).Cordia africanaandO. europaeawere collected from six nurseries;J. procerafrom f ive;D. abyssinica,H. abyssinicaandP. falcatusfrom two; andA. abyssinica,M. ferrugineaandP. africanafrom one nursery. For each of the collected seedlings, age, pot height (PHI), pot diameter (PD), seedling height (SHI) and root collar diameter (CD) were measured and recorded (Table S2). Furthermore, general information such as the name of the nursery,location, geographic coordinates, elevation, potting substrate ratio, ownership data, and names of all tree and shrub species being raised were recorded (Table 1; Table S1). Climate data relevant to each nursery were also gathered from online resources (Table 1).

Root AMF colonization determination

Eight seedlings per species were sampled for root AMF colonization (RC) quantif ication. This was similar to the number of seedlings sampled by Michelsen ( 1992). At the f ield, plastic pots were carefully cut and each seedling was soaked in water to gently detach the root system from the soil. Then, the roots were carefully washed with tap water and f ine roots were cut into 1 cm-long pieces. These f ine roots were mixed and preserved in 50% ethanol in tightly sealed vials for transport to Addis Ababa University (AAU)ecology and ecophysiology laboratory where they were kept at room temperature (18 −23 °C in our case) until RC was determined within one month.

Fig. 1 Outline map of Ethiopia,showing locations of nurseries sampled for this study. N1:Susuni, N2: Entoto, N3: Teferi and Mulugeta, N4: Sokondo,N5: Jello, N6: Qebenewa, N7:Debrebirhan teachers college,N8: Mush, N9: Tis aba lima,N10: Hizbateklehaimanot

Root colonization was determined on three replicates of 2−3 g f ine root subsamples per species and for 200 gridline interaction points (Giovanetti and Mosse 1980). A grid of 2.54 × 2.54 cm was f irst marked beneath the 90 mm diameter plastic Petri dish with a marker. AMF staining was performed by the root staining and de-staining technique using ink and vinegar, based on Vierheilig et al. ( 1998). Before that, roots were cleaned and then cleared. To clean the preserved roots from alcohol, they were thoroughly washed with tap water. To clear the roots, they were autoclaved for 10 min in 10% KOH in tightly sealed vials (Brundrett et al.1996). ClearingJ. proceraseedlings roots was not possible by using 10% KOH or by using 10% H2O2or 10% household bleach after 10% KOH. Soaking the roots in 10% KOH for ten days at room temperature did not clear them either. The clearing was only possible by using the modif ied alkaline hydrogen peroxide method described by Brundrett et al.( 1996). Hence, without the use of 10% KOH,J. proceraroots were directly soaked in 0.50% NH 4 + 0.50% H2O2 (1:1 volume) and were left on a shaker at medium speed for 20 h.Once the roots were cleared, they were stained by soaking them in 5% ink (Hero black ink, made in China) in white vinegar at room temperature for 20 h. They were then destained by rinsing the stained roots in tap water acidif ied with a drop of vinegar for a minimum of 20 min and further rinsing it in tap water until RC was determined after a few minutes or hours using a dissecting microscope at 5.5 × magnif ication. Roots were considered to be colonized with AMF if stained structures of arbuscules, vesicles, aseptate hyphae(extraradical and intraradical mycelia) and/or intraradical spores were observed (S3).

Black Hero ink was used for AMF staining because it was found to give very good contrast (Fig. 2). At the time of this investigation, the only fountain pen inks that were found in the Ethiopian market were permanent black Parker-Quink (France), washable blue Hero (China) and permanent black Hero (China) inks. Hence, these three inks were compared usingO. europaeaseedlings root. Black Hero ink was found to be most suitable (Fig. 2). Since, Trypan Blue dye,commonly used to stain AMF structures of roots by most researchers in Ethiopia and worldwide, is now considered to be carcinogenic (Vierheilig et al. 2005), our use of ink is justif ied. Besides, Trypan Blue is much more expensive and is rarely found in Ethiopian markets.

Spore abundance determination

Pot soil composition Owner GO GO Private GO GO GO NGO GO GO GO 3:1:1 (TS:S:C)4:1:1 (TS:S:C)10:5:3 (TS:M:S)3:2:1 (FS:C:S)12:1:1 (FS:S:C)10:5:1 (FS:C:S)3:1:1 (TS:S:C)3:2:1 (FS:C:S)2:2:1(FS:S:C)3:1:1(FS:S:C)MAT (°C) a MAR (mm) a Köppen climate class b Monsoon-infl uenced temperate oceanic (Cwb)Monsoon-infl uenced temperate oceanic (Cwb)Warm-summer Mediterra- nean (Csb)Warm-summer Mediterra- nean (Csb)Tropical savannah (Aw)Tropical savannah (Aw)Monsoon-infl uenced temperate oceanic (Cwb)Monsoon-infl uenced temperate oceanic (Cwb)Monsoon-infl uenced humid subtropical (Cwa)Hot semi-arid (BSh)1235.9 1235.9 1449.4 1623.4 843.4 843.4 1157.7 1157.7 1030.3 600.3 Climate 15.7 15.7 16.3 17.1 23.2 23.2 16.3 16.3 18.9 23.3 Geographic coordinate Elevation 2515 East 39.67225° 2945 North 9.037944° 38.816°9.080361° 38.75219° 2687 8.958833° 37.75744° 2028 9.042833° 37.42175° 2309 9.030611° 40.85044° 2171 9.011139° 40.86075° 2509 11.4605° 39.64417° 1493 Table 1 Description of the ten DAF nurseries surveyed Location Addis Ababa Addis Abeba Tokae kutay/west shewa Cheleya, west shewa Chiro, West hararge Chiro, West hararge Bosenawerana, north shewa 9.672333° 39.521391° 2759 Bosenawerana, north shewa 9.773°Ambassel, south wollo Enda mehoni, south Tigray 12.79417° 39.55339° 2341 Code Name of the nursery Susuni Entoto Teferi and Mulugeta Sokondo Jello Qebenewa Debre birhan teachers col- lege Mush Tis aba lima N1 N2 N3 N4 N5 N6 N7 N8 N9 N10 Hizba teklehaimanot MAT mean annual temperature, MAR mean annual rainfall, GO Government organization, NGO non-government organization, FS forest soil, TS topsoil, S sand, C compost, M manure a Source: https://clima techa rts.net (accessed 02.08.19); based on 1987−2016 weather data b Source: https://en.wikip edia.org/wiki/K%C3%B6ppe n_clima te_class ifi ca tion (accessed 02.08.19)

Fig. 2 Roots of Olea europaea subsp. cuspidata with arbuscular mycorrhizal fungi structures stained using (1) washable blue Hero ink, (2) black Parker-Quink ink, (3) black Hero ink.Black Hero ink resulted in better contrast where a indicates root not colonized with AMF and b indicates root colonized with AMF. Note: all pictures have similar scales

The remaining eight seedlings collected per species were sampled for spore abundance (SA) quantif ication. In the f ield, seedling shoots were cut and plastic pots removed,and the soil containing the roots was placed in plastic bags(eight seedlings per bag), gently broken, mixed and transported to AAU. At AAU, the soil was allowed to air dry for a few days and then was thoroughly mixed to form one composite sample. SA was determined in three replicates by taking 50 g subsamples from each composite soil sample. Spores were extracted from soil by wet-sieving(1 mm, 180 μm, 90 μm, and 53 μm sieve sizes) followed by density gradient centrifugation in 50% sucrose (Brundrett et al. 1996). Extracted spores were counted on a 90 mm plastic Petri dish according to INVAM protocol ( https://invam .wvu.edu/metho ds/spore s/enume ratio n-of-spore s [accessed 10.01.19]) using a dissecting microscope at 2 × magnif ication. SA in spore number per 50 g soil sample was computed from the average spore numbers of 40 random f ields of observations per Petri dish. The ocular f ield diameter of the microscope was determined to be 9 mm and hence, 100 observations were needed to cover the 90 mm Petri dish. Rarely, spores covered with soil,clusters of spores and sporocarps were observed and were also counted. SA values per 50 g were f inally converted to SA (g −1 ).

Data analysis

One-way ANOVA was computed to determine the Effect of tree species and nursery location (a proxy variable to nursery management) on RC and SA. This was done by considering all the tree species and considering the commonly raised tree species;C. africana,J. procera, andO.europaeaseparately. We used SPSS version 20.0 statistical software to check data for the normality of residuals and equality of variances. Parametric ANOVA was computed when residuals were normal and variances equal. Parametric ANOVA was also computed when data rectif ication by log10 transformation was possible (Effect of nursery location on SA). In other cases, the Kruskal-Wallis test was computed.When Effects proved signif icant (P< 0.05), post-hoc tests,viz. Tukey-Kramer (P< 0.05), Tukey-HSD (P< 0.05) and Dunn-Bonferroni (P< 0.05), were carried out after parametric ANOVA with unequal sample sizes, parametric ANOVA with equal sample sizes, and Kruskal-Wallis tests. We evaluated relationships between RC, SA, seedling age, pot height(PHI), pot diameter (PD), pot volume (PV), seedling height(SHI), seedling collar diameter (CD), SHI per age (RSHI),and CD per age (RCD). Spearman’s rank correlation was computed for all tree species andC. africana,J. procera,andO. europaeaseparately. The scatter plots and bar graphs used to display some of the results were plotted using statistical software SYSTAT version 13.0.

Results

Overview of DAF nurseries

Almost all of the nurseries surveyed were owned by government organizations (Table 1) and, except N3, they were wellstaff ed with manpower and equipment. Based on discussions with forestry offi cials representing districts in our study area,we found that many DAF nurseries no longer raised native trees. Surveyed nurseries raised higher proportions of exotic tree species compared to native species (around 40% native tree species-Table S1). Compared with all the tree and shrub species raised in the surveyed nurseries, the native treesC.africana,J. proceraandO. europaea, were among the most widely raised whileA. abyssinica,M. ferruginea,P. falcatus,andP. africanawere least frequent (Table S1).

Root AMF colonization and pot soil spore abundance

Average RC levels ranged from 8.00 to 99.67% while average SA levels ranged from 1 to 25 spores g −1 . Based on the RC rating adopted by Michelsen ( 1992), (i.e., low = 0−20%,medium = 21−50% and high = 51−100%), 53.8%, 30.7%,and 8.3% of the samples had high, medium and low rates of RC, respectively, while based on our SA rating (i.e.,low = 0−5, medium = 6−10, high = 11−20, and very high = above 20 spores g −1 ), 3.80%, 30.80%, 46.10%, and 19.30% of the samples had very high, high, medium and low SA rates, respectively (Table 2).

Across nurseries,O. europaeaseedlings had high rates of RC regardless of the SA level.Cordia africanaseedlings also mostly had high RC regardless of the SA levels.Juniperus proceraseedlings had medium to low RC levels.Acacia abyssinicahad high RC and very high SA. None of the seedlings had simultaneously low RC and SA rates.However,J. procerafrom N7 and N10,P. falcatusfrom N3 andP. africanafrom N2 (collected only in that nursery) had the least RC and SA rates (Fig. 3). RC and SA rates generally were not correlated (Fig. 3).

Tree species and location Effect on RC and SA

Six of the nine tree species and eight of the ten nurseries were suitable for the ANOVA test. Those species collected only once and those nurseries from which a single species was collected were not suitable for statistical analysis. Due to this data mismatch, two-way ANOVA was not computed and thus, interaction Effects were not determined. Parametric and non-parametric one-way ANOVA tests showed that RC was aff ected by tree species but not by nursery location. In contrast, SA was found to be aff ected by nursery location and not by tree species (Table 4).

The mean RC ofD. abyssinicawas the highest compared to the mean RC of the six tree species analyzed(Fig. 4). Moreover, it was signif icantly, 146.72%, 273.78%,and 316.7% higher than the mean RC ofH. abyssinca,J.procera, andP. falcatusrespectively. On the other hand,the highest mean SA form those nurseries considered for analysis, was found at N1. The mean SA at N1 was also signif icantly, 338.00% and 517.62% higher than the mean SA at N10 and N8 respectively (Fig. 4).

To determine the Effect of nursery management, we evaluated the Effect of location on RC and SA of the three most commonly raised native tree species;C. africana,J.proceraandO. europaea. Nursery location had a signif icant Effect on these three species (Table 5). The mean RC ofC. africanaseedlings was the highest at N5 and it was signif icantly, 218% higher than the mean RC at N2. The highest mean SA ofC. africanaseedlings was recorded at N1 which was signif icantly, 110.00%, 322.00%, and 564.69% greater than the mean RC at N5, N8 and N10,respectively. In the case ofJ. procera, the highest mean RC was found at N2 and with 420.87% more colonization,it was signif icantly higher than the mean RC at N7 only.On the other hand, the mean SA ofJ. proceraseedlings was the highest at N1. Moreover, the mean SA at N1 was signif icantly higher than the mean values at N7 and N10 with the corresponding comparative abundance increases of 86.50% and 90.70%.Olea europaeaseedlings had the highest mean RC at N6. The mean RC at N6 was also signif icantly, 9.57%, 23.00%, 42.60%, and 46.46% greater than the mean RC at N8, N10, N9, and N5 respectively.Hence, mean RC ofO. europaeaseedlings varied markedly even at the very closely located nurseries, N5 and N6.On the other hand, the highest mean SA ofO. europaeaseedlings was found at N2 and was also significantly higher than mean SA at N6, N9, N10, and N8 with the corresponding comparative increases of 45.95%, 573.00%,652.74%, and 1123.28% respectively (Fig. 5).

Table 2 Root AMF colonization (RC) and spore abundance (SA) of the surveyed tree seedlings across nurseries with the corresponding ratings

Correlation

Spearman’s rank correlation results indicated that when all tree species were considered, there was very weak correlation between RC and SA (r s= 0.028,P= 0.891) and between RC and age (r s= − 0.050,P= 0.807). The results also indicated that RC and SA correlations with pot and seedling growth-related variables were weak and not statistically signif icant (Table 6). Analyzing data ofC. africana,J. procera, andO. europaeaseparately, however,most of the correlation results showed a moderate-high strength. Signif icant correlations were those between RC and pot diameter (PD) (r s= 0.820,P= 0.046), between RC and pot volume (PV) (r s= 0.928,P= 0.008), between SA and age (r s= 0.943,P= 0.005), and between SA and root collar diameter per age (RCD) (r s= -0.812,P= 0.050)forC. africana, and between RC and age (r s= 0.829,P= 0.042) forO. europaea(Table 6).

Discussion

The production of resilient native tree/shrub seedlings is fundamental to DAF restoration. This is because passive restoration, i.e., restoration without tree/shrub planting, is hardly possible (Lemenih and Teketay 2004 ; Aerts et al.2007). Therefore, our survey of AMF status in DAF nurseries is relevant. Our results indicate that, similar to a report from a tropical forest in Brazil (Zangaro et al. 2000), RC levels varied between tree functional groups. RC levels of the early-mid successional tree species;A. abyssinica,C.africana,D. abyssinica,andO. europaeawere high while the mid-late successional tree species, viz.J. procera,P.falcatus,andP. africanahad medium or low RC rates. Compared with another study from other DAF nurseries in Ethiopia (Michelsen 1992), RC levels recorded in our study were similar forP. falcatusbut were higher forA. abssynica, C.africana,J. procera,andO. europaea.

Fig. 3 Rates of AMF root colonization (RC) and spore abundance(SA) for all tree species and across the nurseries. Acac.abys Acacia abyssinica, Cord.afri Cordia africana, Dovy.abys Dovyalis abyssinica, Hage.abys Hagenia abyssinica, Juni.proc Juniperus procera,Mill.ferr Millettia ferruginea, Olea.euro Olea europaea subsp Cuspidata, Podo.falc Podocarpus falcatus, and Prun.afri Prunus africana.N1: Susuni, N2: Entoto, N3: Teferi and Mulugeta, N4: Sokondo, N5:Jello, N6: Qebenewa, N7: Debre birhan teachers college, N8: Mush,N9: Tis aba lima, and N10: Hizba teklehaimanot. A: High RC rate,B: Medium RC rate, C: Low RC rate, 1: Low SA rate, 2: Medium SA rate, 3: High SA rate, 4: Very high SA rate. None of the seedlings fall under the (C, 1) category indicating none of them had both low RC and SA rates

Wubet et al. ( 2009) reported that the AMF species composition ofO. europaea,P. falcatusandP. africanaseedlings and their conspecif ic adult trees varied signif icantly.Similarly, our results indicated that RC levels also varied between seedlings and conspecif ic adult trees of mid-late successional tree species. Hence, the RC levels we recorded for the mid-late successional tree species (J. procera,P. falcatus,andP. africana) were lower than those previously reported for their conspecif ic adult trees in DAF (Wubet et al. 2003a; Birhane et al. 2017). However, in the case of the early-mid successional tree species (C. africana,D. abyssinica, andO. europaea), RC levels we recorded were comparable with those reported for conspecif ic adult trees (Wubet et al. 2003a; Birhane et al. 2017, 2018). Hence, mid-late successional tree species could be less responsive to AMF inoculation compared to early-mid successional tree species at the seedling stage (Zangaro et al. 2003). Carrillo-Garcial et al. ( 1999) argued that seedlings of late-successional tree species could maintain low RC levels for the f irst year but become highly infected afterward. If proven true, this may indicate that AMF inoculation of mid-late successional tree species would be more relevant not in the nursery but during f ield plantation after seedlings grew for more than a year.

The SA levels we recorded were comparable to those reported by Birhane et al. ( 2018), but lower than those reported by Delelegn et al. ( 2017) and the SA levels reported within and around the DAF. Seedlings from N1 and N4 had higher SA than those from N8 and N10. This diff erence was probably not due to pot substrate ratios or climate diff erences between these nurseries because neither of these variables showed directional conforming to trends seen among the high-to-low SA nurseries. Therefore, this diff erence was most likely related to the type of potting substrate. We observed that the potting soil at N1 and N4 was red clay soil while at N8 and N10 it was black loam soil. Silva-Flores et al. ( 2019)reported that SA could be signif icantly inf luenced by clay content of potting soils. The diff erence might also be related to other nursery management or soil physicochemical factors(Apple et al. 2005; de Oliveira and de Oliveira 2010; Birhane et al. 2018) not considered in this study.

Generally, no seedlings had both low RC and SA at the same time, hence, inoculation of DAF seedlings in the nursery or during f ield planting may be less important than thought. However,J. procera,P. falcatus,andP. africanahad the lowest levels of both RC and SA, and these could be considered for inoculation. We found RC to be mainly aff ected by host species while SA was aff ected by nursery location. This corroborates f indings reported by John ( 1980),Schüßler et al. ( 2016), and Silva-Flores ( 2019). John ( 1980)reported that RC levels of a host species might fall within a given range for that species. Silva-Flores ( 2019) on the other hand, found that SA was aff ected by soil factors but not by host species. A nursery AMF inoculation experiment by Schüßler et al. ( 2016) showed that RC was moresignif icantly inf luenced by host and AMF species than by soil factors. The fact thatA. abyssinicaandM. ferruginea(Fabaceae) both had high RC rates whileH. abyssinicaandP. africana(Rosaceae) had variable RC rates, also agrees with the assertion by John (1980) that RC levels at the plant genus and family levels could show variable trends.

Table 4 Parametric one-way ANOVA and Kruskal-Wallis test results for the Effect of tree species and nursery location on seedlings root colonization (RC)and spore abundance (SA)

Fig. 4 a mean percent root colonization (RC) of Cordia africana(Cord.afri), Dovyalis abyssinica (Dovy.Abys), Hagenia abyssinica(Hage.Abys), Juniperus procera (Juni.proc), Olea europaea subsp.cuspidata (Olea.euro), Podocarpus falcatus (Podo.afri) in percent and b mean spore abundance (SA) in spore number g− 1 across nurseries.N1: Susuni, N10: Hizba teklehaimanot, N2: Entoto, N4: Sokondo,N5: Jello, N6: Qebenewa, N7: Debre birhan teachers college, N8:Mush. Diff erent letters indicate signif icant diff erences after Post-hoc Tukey tests ( P < 0.05) for the Effect of a species and b location

Table 5 Parametric ANOVA and Kruskal-Wallis test results for the Effect of location (proxy for nursery management) on root colonization (RC) and spore abundance (SA) of seedlings by species

Root colonization and SA ofC. africana,J. procera, andO. europaeasignif icantly varied across nurseries indicating that nursery management had inf luenced both factors.ForJ. proceraandO. europaea, nurseries that resulted in low RC and high RC levels corresponded to those raising younger and older seedlings, respectively. ForO. europaeain particular, the RC level at N6 was signif icantly highest and at N5, it was lowest. These nurseries were close to one another with similar climate and maybe, comparable soil properties. They diff ered in pot substrate ratio and age of seedlings, viz. at N6, they were 11 months in age while at N5, they were 5 months. This may indicate that age had a signif icant Effect on RC ofO. europaeaseedlings. The correlation results also indicated that the RC-age correlation forO. europaea, was signif icant, very strong and positive (r s= 0.829,P= 0.042). Similarly Michelsen( 1992) reported that younger nursery seedlings tend to have lower RC than older seedlings. According to our results,the Effect of seedling age was varied amongC. africana,J. procera, andO. europaea. Moreover, RC-age correlation with all tree species was insignif icant (r s= − 0.050,P= 0.807), indicating RC-age correlation was most probably host-dependent. Similarly, Abbott and Robson ( 1991)compiled three diff erent works that showed age and RC correlations can vary widely between diff erent host species.

The signif icant very strong positive RC-pot diameter (PD)and RC-pot volume (PV) correlations we recorded forC.africanasupport previous reports although for other plant species (Audet and Charest 2010; Zangaro et al. 2015).The strong negative SA−RSHI and the signif icant negative SA−RCD correlations forC. africanaand the strong negative RC−RSHI and RC−RCD correlations forO. europaeacould be related to AMF parasitism as suggested by Johnson et al. ( 1997). However, the most probable reason for the observed negative correlations could be pseudo negative growth rates due to small pot size. Due to smaller pot height and diameter,C. africanaandO. europaeaseedlings might not grow to their physiological potential. Hence, seedlings could remain with the same SHI and CD despite the increase in age (i.e., negative growth rates) while SA and RC on the other hand, increased with age.

Fig. 5 Mean root colonization(RC) in percent and spore abundance (SA) in spore number g− 1 of the three common tree species; Cordia africana ( C.africana), Juniperus procera( J. procera) and Olea europaea subsp. cuspidata ( O. europaea)across nurseries. N1: Susuni,N2: Entoto, N4: Sokondo, N5:Jello, N6: Qebenewa, N7: Debre birhan teachers college, N8:Mush, N9: Tis aba lima, N10:Hizba teklehaimanot. Diff erent letters indicate signif icant diff erences after Post-hoc Tukey and Dunn-Bonferroni tests ( P < 0.05) for the Effect of location

The weak correlation we recorded between RC and SA corroborates the results of Abbott and Robson ( 1982) and Douds and Schenck ( 1990). The fact that AMF community composition of roots, spores and extraradical mycelia has been shown to vary (Varela-Cervero et al. 2015) also indicates that RC and SA could be independent. Moreover, the RC-SA correlation could depend on AMF species composition in the rhizosphere,with diff erent AMF species requiring diff erent critical RC levels to reach maximum sporulation (Gazey et al. 1992).

Despite its importance in planning AMF inoculation(Verbruggen et al. 2013), in this study we did not describe the AMF composition of seedlings. We recommend that future nursery research should consider the AMF composition of seedlings to better inform AMF inoculation projects.We also did not quantify soil physicochemical properties because these variables may be less likely to be modif ied by nursery managers. However, future research could consider these variables as they potentially aff ect RC and SA. In the previous DAF nursery AMF survey, SA was not quantif ied by tree species or by nursery. Considering both RC and SA in AMF nursery surveys is important since the root, spores,and external hyphae are all sources of various AMF inocula(Varela-Cervero et al. 2015). It is also important becausediff erent AMF species have diff erent modes of infection(Hart and Reader 2004; Chagnon et al. 2013) and AMF species favored by trees may diff er at various growth stages(Husband et al. 2002). We quantif ied both RC and SA per individual seedling across nurseries. Future nursery surveys could further consider soil hyphae abundance such that all AMF inoculum pools are evaluated.

Table 6 Spearman’s rank correlation [ P (2-tailed)] of root colonization (RC), spore abundance (SA) and various pot and seedling growth-related variables

To our knowledge, this is the f irst large scale AMF study conducted by using black Hero ink for AMF staining. The suitability of black Hero ink that we experienced is partly in agreement with Vierheilig et al. ( 2005), who concluded that almost any black ink could be suitable for AMF staining.However, it contrasts with the f indings of Cao et al. ( 2013)who concluded that washable blue Hero ink was more suitable for AMF staining.Acacia abyssinica,M. ferrugineaand most importantly,D. abyssinicaseedling roots were very highly infected with AMF and hence, these species could be potential trap-tree species to prepare root AMF inocula in the future. Since root inoculum can be produced in a short period and since it can infect seedling roots much faster,root inoculum is considered to be a preferred alternative to conventional crude inoculum (Habte and Osorio 2001).

Conclusion

Our study of the AMF status (AMF root colonization and spore abundance) of ten DAF nurseries in Ethiopia is the second of its kind in Ethiopia. We have, maybe for the f irst time, demonstrated that the locally available black Hero ink proved suitable for AMF root colonization (RC) determination. Hence, future AMF surveys at nursery or f ield level could consider it as an alternative stain.

The RC rates we recorded for seedlings at DAF nurseries were, contrary to the previous report, mostly high. Moreover,none of the seedlings had low AMF inoculum, i.e., simultaneously low RC and SA. We produced little evidence to suggest AMF inoculation of DAF seedlings in the nursery or f ield planting was needed. However,J. procera,P. falcatus, andP.africanaseedlings had low levels of AMF inoculum, and these species might benef it from supplemental AMF inoculation.The fact that RC rates in roots of early-mid successional and mid-late successional tree species varied markedly with the former having high RC rates could indicate that mid-late successional tree species may be less responsive to AMF inoculation at seedling stage and if inoculation is prescribed, it should be completed during f ield planting. However, prior to recommending for or against AMF inoculation, it would be useful if AMF composition of seedlings and planting sites is known. In future, AMF species composition of colonized roots and pot soils at various DAF nurseries and for diff erent tree functional groups should be documented.

RC was mainly aff ected by host species while SA was aff ected by nursery location (and possibly by soil structure and/or chemistry). Both SA and RC of an individual tree species were aff ected by nursery location (most probably seedling age and pot volume). These indicate that although host species and soil factors may be the main variables aff ecting RC and SA respectively, there could as well be two-way or three-way interactions between host species, soil factors and nursery management variables such as age and pot volume or other variables not considered in this study. Our f inding of the diff erential main Effects that host species and nursery location (soil factors) have on RC and SA is important in terms of AMF ecology and needs further investigation.

Our results indicate that, when all tree species were considered, none of the nursery management or seedling growth variables was correlated with RC or SA. There was also no correlation between RC and SA or between RC and seedling age. However, when individual tree species were considered,age and pot volume were correlated forO. europaeaandC. africanarespectively. Hence, nursery managers could be advised not to takeO. europaeaseedlings out for plantation at a very young age despite height attainment. Nursery managers could also use larger volume pots to improve AMF inoculum ofC. africanaseedlings. However, additional experiments are needed to further elucidate the Effects of age and pot volume on diff erent tree functional groups.

We sampledP. falcatusandP. africanafrom only two and one nurseries, respectively. Hence, future studies should sample these species from more nurseries. This will be possible if more DAF nurseries start to raise these species. Our results, similar to other previous studies, indicate that AMF status can vary between seedlings and conspecif ic adult trees, especially for mid-late successional tree species.Therefore, future nursery-level AMF studies should be conducted to complement f ield AMF studies such that the AMF role in forest restoration is better understood.

Acknowledgements The Addis Ababa University (AAU) and Ethiopian Biodiversity Institute (EBI) are acknowledged for providing the fund to carry out this research which is part of the Ph.D. work by the f irst author. We would like to thank Addis Ababa city administration environment protection authority and the agricultural bureaus at the zones and districts where the investigated nurseries were located for providing the seedlings and useful information. We also thank sincerely the two anonymous reviewers for their relevant constructive comments on the earlier draft.

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affi liations.