Arbuscular mycorrhizal fungi associated with Populus–Salix stands in a semiarid riparian ecosystem


Author for correspondence: Vanessa Beauchamp Tel: +970 226 9381 Fax: +970 226 9230 Email:


  • • This study examined the activity, species richness, and species composition of the arbuscular mycorrhizal fungal (AMF) community of Populus–Salix stands on the Verde River (Arizona, USA), quantified patterns of AMF richness and colonization along complex floodplain gradients, and identified environmental variables responsible for structuring the AMF community.
  • • Samples from 61 Populus–Salix stands were analyzed for AMF and herbaceous composition, AMF colonization, gravimetric soil moisture, soil texture, per cent organic matter, pH, and concentrations of nitrate, bicarbonate phosphorus and exchangeable potassium.
  • • AMF species richness declined with stand age and distance from and elevation above the channel and was positively related to perennial species cover and richness and gravimetric soil moisture. Distance from and elevation above the active channel, forest age, annual species cover, perennial species richness, and exchangeable potassium concentration all played a role in structuring the AMF community in this riparian area.
  • • Most AMF species were found across a wide range of soil conditions, but a subset of species tended to occur more often in hydric areas. This group of riparian affiliate AMF species includes several not previously encountered in the surrounding Sonoran desert.


Arbuscular mycorrhizal fungi (AMF) are soil fungi that form symbiotic associations with the roots of many plant species and are important in plant nutrition. Individual fungal species are able to interact with many plant species (Allen, 1991; Smith & Read, 1997) and fungal communities are perceived as being depauperate in comparison to most plant communities (Allen et al., 1995). Only c. 250 species of AMF have been described, and until recently these fungi were assumed to be functionally redundant (Bever et al., 2001). However, recent research has shown that some AMF species are ecologically distinct in their phenology (Gemma et al., 1989; Lee & Koske, 1994; Schultz et al., 1999; Pringle & Bever, 2002), their distribution with respect to edaphic factors (Wetzel & van der Valk, 1996) including soil pH (Abbott & Robson, 1977; Porter et al., 1987), moisture (Anderson et al., 1984; Rickerl et al., 1994; Miller & Bever, 1999; Miller, 2000; He et al., 2002) and nutrients (Anderson et al., 1984; Johnson et al., 1991, 1992; He et al., 2002), and their effect on plant species (Streitwolf-Engel et al., 1997, 2001; van der Heijden et al., 1998a).

The composition of AMF communities and their relationship to environmental gradients has been investigated in old-field (Johnson et al., 1991; Barni & Siniscalco, 2000), desert (Stahl & Christensen, 1982; Bethlenfalvay et al., 1984; Stutz et al., 2000), grassland (Gibson & Hetrick, 1988; Eom et al., 2001; Eriksson, 2001; Landis et al., 2004), tropical (Brundrett et al., 1996; Johnson & Wedin, 1997; Lovelock et al., 2003), urban (Stabler et al., 2001; Cousins et al., 2003), and wetland (Anderson et al., 1984; Turner & Friese, 1998; Miller & Bever, 1999; Turner et al., 2000) environments; however, little is known about the composition or distribution of the AMF community associated with riparian ecosystems in arid or semiarid regions (Kennedy et al., 2002). In the south-western United States, riparian ecosystems dominated by Populus and Salix form ecotones between aquatic habitats associated with lakes and rivers, and dry upland deserts (Gregory et al., 1991; Naiman & Decamps, 1997). These oases comprise only 1% of the land area in the south-west but are vital as wildlife habitat and migration corridors (Gregory et al., 1991; Briggs, 1996). These areas contain a unique assemblage of plant species not found in permanently flooded wetlands, nor in the arid deserts (Szaro, 1990; Minckley & Brown, 1994), and may also contain unique AMF species.

Wetlands usually have strong gradients of water depth, soil moisture and nutrient availability (Mitsch & Gosselink, 2000) and, although AMF colonization does occur in flooded environments, most measures of AMF activity and abundance, including root colonization, spore numbers and species richness, are negatively correlated with soil moisture or water depth (Anderson et al., 1984; van Duin et al., 1990; Rickerl et al., 1994; Stevens & Peterson, 1996; Turner & Friese, 1998). Studies in deserts show the opposite trend, with AMF activity and species richness positively correlated with soil moisture (Zak et al., 1995; Jacobson, 1997; He et al., 2002). Environmental gradients in riparian ecosystems include decreases in soil moisture with increasing distance from and elevation above the active channel (Boggs & Weaver, 1994). Investigations in riparian Sporobolus wrightii Munro ex Scribn (big sacaton) grasslands along the San Pedro River in south-eastern Arizona found a positive correlation between AMF colonization and soil moisture (Kennedy et al., 2002), mimicking patterns of the desert upland. It is unknown if AMF communities in PopulusSalix-dominated floodplains exhibit activity and distribution patterns similar to those of wetlands or the arid desert. Other environmental gradients in riparian systems that could influence the activity, richness and distribution of AMF include increases in the percentage of organic matter, the concentrations of phosphorus, nitrogen and potassium (Boggs & Weaver, 1994) and herbaceous richness (Menges, 1986; Lite et al., 2005) as stands age and elevation above and distance from the channel increase.

Understanding how the AMF community is structured with respect to environmental gradients in riparian ecosystems is an important step towards understanding the function of AMF within this ecosystem and the ecological requirements of AMF species. Additionally, information about the species composition and distribution of the AMF community in PopulusSalix forests may enhance the success of efforts to restore degraded riparian areas. Populus and Salix are capable of forming relationships with both AMF and ectomycorrhizal fungi (Lodge, 1989; van der Heijden & Vosatka, 1999) and both types of mycorrhizal fungi are important and little-studied components of riparian areas. However, in this study we chose to focus only on the AMF associated with Populus–Salix stands, as recent work by Jacobson (2004) has begun to examine the ectomycorrhizal fungi present in these systems but no other researchers, to our knowledge, are investigating AMF.

The objectives of this study were: to identify AMF species associated with south-western riparian PopulusSalix forests; to identify environmental variables related to AMF species richness and community composition in a Sonoran Desert riparian area; and to examine AMF community patterns along complex environmental gradients in river floodplains. We expected that AMF activity and richness would change along a successional gradient of Populus–Salix stands and that these changes would be related to changes in soil or vegetation that occur as the stands age. We anticipated that the AMF activity and composition of the youngest Populus–Salix stands would be most different from those of the older age classes, either because AMF activity would be restricted in young stands by the anoxic conditions created by flooding (Anderson et al., 1984; Rickerl et al., 1994) or because it would be stimulated by the high levels of soil moisture maintained by stream flow and shallow groundwater levels (Zak et al., 1995; Jacobson, 1997; He et al., 2002).

In this study, we used AMF spore wall characteristics to identify the AMF encountered in our surveys of Populus–Salix stands. Molecular identification methods are common in studies of ectomycorrhizal fungi (Dahlberg, 2001; Horton & Bruns, 2001) and are being increasingly used in the study of AMF (Clapp et al., 1995; Vandenkoornhuyse et al., 2002, 2003; Douhan et al., 2005; Stukenbrock & Rosendahl, 2005). These methods have the potential to revolutionize studies of AMF community structure, but their use is hampered by several obstacles including the polymorphic, multigenomic nature of AMF, the lack of a single primer to amplify AMF DNA and the difficulty in matching sequence groups to morphospecies (Redecker et al., 2003). Additionally, the potentially large numbers of AMF species that may be encountered in ecological studies across broad environmental gradients make the use of molecular methods prohibitively time consuming and expensive (Landis et al., 2004). While extensive molecular analysis was outside the scope of this study, the identification and isolation of AMF morphospecies from semiarid riparian areas should aid in the use of molecular identification techniques in future studies.

Materials and Methods

Study area

The Verde River originates in the central Arizona highlands near Paulden, Arizona, and flows east toward Camp Verde and then south to its confluence with the Salt River north-east of Phoenix, Arizona (Owen-Joyce & Bell, 1983). The river flows freely for approx. 125 km and then is impounded by two dams operated for water storage 60 km north of Phoenix (Fig. 1). High flows in the Verde typically occur in early spring and are driven by winter Pacific frontal storms and by snowmelt in the upper elevations of the watershed (Owen-Joyce & Bell, 1983). These floods occur infrequently (once every 5–10 years) and create conditions necessary for recruitment of pioneer woody riparian trees such as Populus and Salix (Stromberg et al., 1991, 1993; Stromberg, 1993). Establishment of riparian trees along stream channels and subsequent channel movement through channel accretion and migration create a mosaic of habitats on the floodplain, with younger trees typically nearer the active channel and older stands further from and higher than the stream channel and nearer to upland environments.

Figure 1.

Map of the Verde River showing study sites, stream gage and dam locations. The inset map indicates the position of the Verde watershed in the state of Arizona, USA. Study sites: 1, Perkinsville; 2, Tapco; 3, Dead Horse Ranch; 4, Horseshoe Dam; 5, Bartlett Dam; 6, Rio Verde; 7, Fort McDowell.

Study design

Sixty-one 100-m2 study plots were situated in Populus fremontii S. Wats.–Salix gooddingii Ball stands on the floodplain of the Verde River. These plots were distributed over seven sites and across the age range of stands present at each site. Three study sites (27 plots) were situated in the middle Verde Basin, which experiences unregulated flow (designated the unregulated reach), and four sites (34 plots) were located in the lower Verde, which experiences flows altered by the operation of two water storage dams (designated the regulated reach). Plot elevation above and distance from the active channel were measured with a rod and transit or Global Positioning System (GPS).

The age of trees in each sample plot was estimated by coring or slabbing (depending on size) five of the largest P. fremontii or S. gooddingii trees in each plot. Trees less than 6 cm in basal diameter were slabbed at the base with a saw, and trees over 6 cm in diameter were cored as close to the ground as possible (40–60 cm above ground level). Cores were dried and glued into wooden mounts. Cores and slabs were sanded with successively finer grades of sandpaper (to 900 grit) and annual growth rings were counted using a dissecting microscope (Stromberg, 1998). Minimum plot age was defined as the age of the oldest tree slabbed or cored per plot, and plots were divided into three groups based on tree age. Sapling plots contained trees 1–10 years old, mature plots contained trees 11–55 years old, and old-growth plots contained trees older than 55 years.

Between March and April 2001, soil was collected from five randomly selected points within the 100-m2 study plots. Each 350-cm3 sample was taken to a depth of 20 cm (Virginia et al., 1986) with a split-core soil sampler or a trowel, which was rinsed with 70% ethanol between samples. All five subsamples from each plot were bulked together, sealed in plastic bags and stored at 4°C until analysis.

Herbaceous vegetation was sampled at the time of soil collection. Herbaceous cover, by species, using Daubenmire cover classes was recorded in 1-m2 quadrats located adjacent to each soil sample point (Mueller-Dombois & Ellenberg, 1974). Plot cover by species was calculated as the average cover across all five quadrats. Plants were identified using the method of Kearney & Peebles (1960) and taxonomic treatments recently published in the Journal of the Arizona-Nevada Academy of Sciences. Herbaceous species were classified into functional groups based on life-span (annual/biennial or perennial) (USDA-NRCS, 2004) and membership in typically mycorrhizal or nonmycorrhizal plant families (Gerdemann, 1968; Newman & Reddell, 1987; Supplementary material Table S1).

To determine AMF colonization at the time of sampling, fine roots were separated from each soil sample and fixed in 70% ethanol, cleared in 5% KOH and stained in trypan blue (Koske & Gemma, 1989). Colonization of AMF was assessed with the magnified intersections method (McGonigle et al., 1990). A 250-cm3 subsample of each soil sample was used to establish a trap culture (described below) and the remainder was analyzed for gravimetric soil moisture, soil texture, per cent organic matter, pH, and concentrations of nitrate, Laboratory Consultants Ltd (Lordsburg, NM, USA).

Trap cultures were established using collected soil in a glasshouse at Arizona State University (Stutz & Morton, 1996). Field soil was mixed in a 1.25 : 1 ratio with autoclaved #12 and #20 [1 : 1 volume/volume (v/v)] silica sand and placed in a surface-sterilized 650-ml deepot (D40; Stuewe & Sons, Corvallis, OR, USA). Each pot was seeded with approx. 75 seeds of Sorghum sudanese (Piper) Staph (Sudan grass). Trap cultures were grown in the glasshouse for 12 wk, top-watered to flow through daily, and fertilized monthly (Peter's No-Phos fertilizer solution; Scotts-Sierra Horticultural Products Co. 25-0-25; Marysville, OH, USA). After 3 months, irrigation was discontinued and the cultures were allowed to dry. The contents of each pot were placed in plastic bags and stored at 4°C.

AMF spores were extracted from a 100-cm3 subsample of each culture by wet-sieving and decanting followed by centrifugation in a sucrose density gradient (Daniels & Skipper, 1982). Spores were placed in a Petri dish, separated by morphotype based on spore size and color under a dissecting microscope, and mounted and crushed on microscope slides in Meltzer's reagent (Koske & Tessier, 1983). Spores were identified with light microscopy using spore wall characteristics (Morton, 1988; Morton et al., 1993).

Data analysis

Because of the similarity in AMF species richness and composition between reach types (V. B. Beauchamp et al., unpublished), all analyses regarding AMF species richness or composition were performed on the full data set rather than for unregulated and regulated reaches separately. Comparisons of AMF colonization and species richness between Populus–Salix stand age classes were conducted with Mann–Whitney U-tests. AMF species composition was compared between age classes with the Sorenson similarity coefficient (Magurran, 1988).

Spearman rank correlations were used to determine relationships between environmental variables and AMF colonization and species richness (using a Bonferroni-corrected significance level of P = 0.002 to increase confidence in positive findings; Zar, 1999). Environmental variables were also used in stepwise multiple regression analysis to identify significant explanatory variables of AMF species richness. Stepwise regression was performed twice, once using the full set of independent variables (successional, positional, biotic and soil) and again using only soil variables. Stepwise regression used a forward selection model with a significance of P < 0.150 for entry of variables into the model (SAS Institute, 2001).

To evaluate relationships between AMF community composition and environmental variables, we first ordinated the AMF presence/absence data using nonmetric multidimensional scaling (NMDS) and then looked for significant correlations between the ordination axes and environmental variables (McCune & Bedford, 1999; McCune & Grace, 2002). The NMDS ordination was performed using a Sorenson distance matrix and the ‘slow and through’ autopilot mode in PC-ORD (v.4.25, MJM Software Design, Gleneden Beach, OR, USA). Binary AMF presence/absence data were transformed with Beals smoothing before analysis. This transformation technique calculates the probability of a species occurring in each plot based on the joint occurrence of that species with other species that are found in that plot. This reduces the large number of zeros common for presence/absence data sets which can result in unstable solutions in NMDS (McCune & Bedford, 1999; McCune & Grace, 2002). Species encountered in only one plot were excluded from the ordination analyses. Environmental variables expressed as a percentage were arc-sin square-root transformed before analysis and all other variables were log-transformed. After transformation a general relativization was performed on all environmental variables to rectify extreme differences in variable values resulting from measurements made at different scales. Correlations between each ordination axis and environmental variables were calculated to identify significant (r2 > 0.20) associations between the composition of the AMF community and environmental variables (McCune & Bedford, 1999; McCune & Grace, 2002).

To examine soil moisture affinity for each AMF species, the 61 study plots were divided into three soil moisture classes based on soil moisture content by weight (dry: 0.1–1.9%, 19 plots; intermediate: 2.0–4.5%, 20 plots; wet 4.6–20%, 22 plots) and the number of occurrences for each AMF species was compared between the soil-moisture classes.


The flora of Populus–Salix stands sampled for this study consisted of 102 plant species and was dominated, in terms of both cover and richness, by annual species and by members of typically mycorrhizal plant families (Table 1).

Table 1.  Richness and cover of plant species grouped by life history and family-level mycorrhizal affinity
Plant typeAverage plot richnessAverage plot cover (%)% of total richness% of total cover
  1. Values for Average plot richness and Average plot cover are means ± 1 standard error

Annual5.8 ± 0.517.6 ± 2.86776
Perennial2.1 ± 0.3 5.4 ± 1.33324
Mycorrhizal6.0 ± 0.516.3 ± 2.47571
Nonmycorrhizal1.9 ± 0.2 6.7 ± 1.52529

AMF colonization levels of plant roots collected from Populus–Salix stands ranged from 0 to 59% (average 22.8 ± 2.1%). Total colonization and arbuscule and vesicle colonization were highest in sapling stands when compared with other age classes (Table 2) but did not show a significant relationship to any of the other environmental variables measured.

Table 2.  Comparison of arbuscular mycorrhizal fungal (AMF) colonization and species richness among Populus–Salix stand age classes on the Verde River (Arizona, USA)
Age classTotal colonizationHyphaeArbusculesVesiclesAMF richness
  1. Values are means ± 1 standard error.

  2. Different letters within columns indicate significant differences (P < 0.05).

Sapling29.0 ± 3.7 a13.2 ± 2.2 a11.8 ± 1.9 a2.3 ± 0.6 a8.6 ± 0.5 a
Mature20.3 ± 3.7 ab11.0 ± 2.1 a 6.8 ± 2.2 b1.9 ± 0.8 ab5.8 ± 0.4 b
Old-growth17.1 ± 3.7 b10.1 ± 2.8 a 6.4 ± 2.1 b0.5 ± 0.5 b5.0 ± 0.5 b

Thirty species (or morphospecies) of AMF were detected in trap cultures of soil collected from PopulusSalix stands along the Verde River. Fourteen of these species belonged to the genus Glomus, 11 to Acaulospora, three to Entrophospora and one each to Paraglomus and Archaeospora. Nine of these species (one Glomus, two Entrophospora and six Acaulospora) are undescribed. Common species (detected in approximately half the plots sampled) included Glomus intraradices Schenck & Smith, Glomus microaggregatum Koske, Gemma & Olexia, Glomus spurcum Pfeiffer, Walker & Bloss, Glomus deserticola Trappe & Menge, Glomus eburneum Kennedy, Stutz & Morton, and Glomus mosseae (Nicol & Gerd.) Gerdemann & Trappe. Six species were encountered in three or fewer plots. AMF were encountered in samples from every plot and plot species richness ranged from 2 to 13 AMF species.

AMF community composition was similar among age classes, with Sorenson similarity scores between age classes ranging from 0.79 to 0.88 (Table 3). Of the 30 AMF species detected, 18 were found in all three age classes and five were found in two age classes. Six of the AMF species encountered were unique to sapling stands and one was unique to mature stands (Fig. 2). AMF species richness was highest in sapling stands (Table 2) and decreased significantly with increasing plot age and distance from and elevation above the active channel. AMF species richness increased with increasing perennial species richness and cover, but not with increasing total species richness and cover or increasing annual species richness and cover (Table 4). AMF species richness was not significantly correlated with any of the soil variables measured.

Table 3.  Sorenson index of similarity for arbuscular mycorrhizal fungal (AMF) species composition of plots on the Verde River (Arizona, USA) by Populus–Salix forest age class
Sapling 0.820.79
Mature  0.88
Figure 2.

Arbuscular mycorrhizal fungal (AMF) species frequency of occurrence in pot cultures of soil collected from PopulusSalix stands on the Verde River (Arizona, USA). Genera: Ac., Acaulospora; Ar., Archaeospora; E., Entrophospora; G., Glomus; P., Paraglomus. AZ112 and AZ123 are previously discovered but undescribed Glomus species. All other numbered species are additional undescribed species encountered in this study.

Table 4.  Spearman rank correlations (rs) of arbuscular mycorrhizal fungal (AMF) species richness with environmental variables on the Verde River (Arizona, USA)
  1. All values shown are P < 0.05 (Bonferroni-corrected P = 0.002).

Herbaceous cover
Herbaceous richness
Herbaceous diversity
Annual cover
Annual richness
Perennial cover0.41
Perennial richness0.46
Cover of typically mycorrhizal species
Richness of typically mycorrhizal species
Cover of typically nonmycorrhizal species
Richness of typically nonmycorrhizal species
Organic matter

Stepwise multiple regression on all independent variables also identified perennial species cover (positive influence) and plot elevation (negative influence) as significant explanatory variables of AMF species richness. Stepwise regression restricted to soil variables identified moisture (positive influence) as the most significant explanatory variable of AMF species richness. The model based on all variables explained substantially more of the variation in AMF species richness than did the model using only soil variables (Table 5).

Table 5.  Multiple regression analysis of environmental variables contributing to arbuscular mycorrhizal fungal (AMF) species richness on the Verde River (Arizona, USA)
All independent variables
Elevation−0.7890.24519.16< 0.0001
Perennial cover 0.0570.038 3.05  0.0859
Intercept 7.888   
Soil variables
Moisture 0.2690.14910.32  0.0021
Intercept 5.460   

Nonmetric multidimensional scaling (NMDS) identified a two-dimensional solution for the AMF species presence/absence data which explained 51 and 34% (cumulative = 85%) of the variation in species presence/absence data (final stress = 17.105; instability = 1.0 × 10−5). Annual species cover, perennial species richness, distance from and elevation above the active channel, plot age and exchangeable potassium concentration showed significant correlations (r2 > 0.20) with sample axis scores, indicating that these variables likely play a role in structuring AMF composition in this riparian area (Fig. 3).

Figure 3.

Nonmetric multidimensional scaling (NMS) ordination diagram for arbuscular mycorrhizal fungal (AMF) species from PopulusSalix stands. Bold font indicates undescribed ‘riparian affiliate’ species encountered in this study. Arrows indicate the direction and magnitude of environmental variables significantly correlated (r2 > 0.20) with ordination axes. Genera: Ac., Acaulospora; Ar., Archaeospora; E., Entrophospora; G., Glomus; P., Paraglomus. AZ112 and AZ123 are previously discovered but undescribed Glomus species. All other numbered species are additional undescribed species encountered in this study.

No AMF species encountered appeared to be restricted to wet or dry plots, although some showed a tendency to occur more frequently in wet plots and a few others tended to occur more frequently in dry plots (Fig. 4). The majority of the undescribed species encountered in the study (five of nine) had 60% or more of their occurrences in wet plots.

Figure 4.

Hydrologic affinity for arbuscular mycorrhizal fungal (AMF) species encountered in pot cultures from PopulusSalix stands. Each column shows the frequency of encounters for a particular species in dry (solid bars), intermediate (shaded bars) and wet (open bars) plots. The numbers at the top of the graph are the total number of detections for each species. Genera: Ac., Acaulospora; Ar., Archaeospora; E., Entrophospora; G., Glomus; P., Paraglomus. AZ112 and AZ123 are previously discovered but undescribed Glomus species. All other numbered species are additional undescribed species encountered in this study.


The AMF community structure of PopulusSalix stands on the Verde River has some unique characteristics but also has many similarities to that of the surrounding Sonoran Desert. Some of the AMF species most frequently encountered in this study (G. intraradices, G. microaggregatum, G. spurcum, G. eburneum and G. mosseae) are also common in the desert uplands (Stutz & Morton, 1996; Stutz et al., 2000). Additionally, all of the species we encountered were small-spored, a characteristic typical of desert environments (Bethlenfalvay et al., 1984; Jacobson, 1997; Stutz et al., 2000) which may be an adaptation to limit moisture loss in arid environments. The two AMF genera not encountered in this study, Gigaspora and Scutellospora, contain species with much larger spore sizes, and are infrequent in desert ecosystems (Bethlenfalvay et al., 1984; Jacobson, 1997; Stutz et al., 2000). While many of the species encountered in the riparian area are also common to the desert uplands, this study has also identified a suite of AMF species that appear to be riparian affiliates. We detected eight undescribed Acaulospora and Entrophospora morphotypes, two of which were quite common, that have not been found in extensive surveys of the Sonoran Desert (Stutz & Morton, 1996; Stutz et al., 2000). We also detected an additional set of species including Glomus luteum Kennedy, Stutz & Morton, Acaulospora morrowiae Spain & Schenck, and Acaulospora delicata Walker, Pfeiffer & Bloss that have been commonly detected in riparian areas of the San Pedro River in south-eastern Arizona (Kennedy et al., 2002), but infrequently in the Sonoran desert.

Because spore densities in desert soils can be low (Stutz & Morton, 1996; Kennedy et al., 2002), trap cultures were used amplify AMF collected in this study and yielded 30 AMF species associated with riparian PopulusSalix stands. Other work with trap cultures has shown that the season, the geographic location of the glasshouse, the host plant, and other factors can influence sporulation of AMF species (Stutz & Morton, 1996; Bever et al., 2001). Subsequent generations of trap cultures can reveal additional cryptic AMF species (Stutz & Morton, 1996; Bever et al., 2001), so it is likely that some AMF species were not detected and are missing from this analysis.

Several interrelated environmental variables appear to be important influences on AMF activity and species richness in this desert riparian ecosystem, including soil moisture, distance from and elevation above the active channel, stand age and perennial plant species richness. Many studies have examined the relationship of AMF activity and richness to soil moisture and have found that, in wetland environments, AMF richness and activity respond negatively to increasing levels of soil moisture (Anderson et al., 1984; van Duin et al., 1990; Rickerl et al., 1994; Stevens & Peterson, 1996; Turner & Friese, 1998) while in deserts or dune systems AMF presence and activity are positively correlated with soil moisture (Zak et al., 1995; Jacobson, 1997; He et al., 2002). Our study, as well as a desert riparian study by Kennedy et al. (2002), found AMF activity and species richness to decrease with increasing distance from or elevation above the active channel, meaning that the sites with the highest AMF activity and richness tended to be in the lower and wetter floodplain areas. Our study also found a direct positive relationship of AMF species richness with soil moisture and found that AMF colonization was highest in the youngest (and typically wettest) Populus–Salix stands. In this respect, AMF activity and richness in desert riparian ecosystems more closely resemble patterns found in arid deserts than in inundated wetlands. Differences in the hydrology of wetlands and arid region floodplains may explain these differences in AMF activity and species richness patterns. In wetlands, surface soils are permanently or seasonally saturated, creating anaerobic conditions that are thought to restrict AMF activity (Khan, 1974; Mosse et al., 1981; LeTacon et al., 1983). In arid region floodplains, localized areas of permanently saturated soil can develop from short-duration floods or rainfall events, but surface soils throughout the floodplain are dry for much of the year. Groundwater in the south-western desert tends to be oxic, and the fast-flowing flood waters may be well aerated, thus allowing more aerobic microbial activity (Turner et al., 2000). Alternating periods of wetting and drying, which are common in riparian zones, can also stimulate AMF community development (Braunberger et al., 1996; Brown & Bledsoe, 1996; Miller, 2000). AMF colonization and community composition can vary temporally (van der Heijden & Vosatka, 1999; Kennedy et al., 2002; Bohrer et al., 2004) as well as spatially, but seasonal changes in AMF activity and AMF community composition within Populus–Salix stands were not investigated in this study.

The above-ground plant community can also have significant effects on the richness of the AMF community, with positive correlations between plant species richness and AMF species richness found in glasshouse studies (Grime et al., 1987; van der Heijden et al., 1998b; van der Heijden, 2004) and field studies (Landis et al., 2004). However, in this study, plot-level measures of total herbaceous cover, richness or diversity were not correlated with AMF species richness. We did find that, when the flora was divided into annual and perennial species, AMF richness was positively correlated with cover and richness of perennial herbaceous species. This differentiation is likely attributable to differences in dominance and mycorrhizal dependence of herbaceous perennials and annual or short-lived ephermeral plants. As in most desert environments, a substantial portion of the riparian flora in arid and semiarid riparian areas has an annual life-span and can have a rapid response to rainfall events (Bagstad et al., 2005); these species can dominate the flora in moderate- to high-rainfall years or wet seasons, such as the year and season of this study. Unlike the herbaceous perennials, these precipitation-reactive species, which are mostly ephermeral or short-lived annual species, are unlikely to be highly dependent on mycorrhizal fungi (Allen, 1991; Collier et al., 2003).

While we did find a relationship between AMF richness and the life history characteristics of the above-ground vegetation, we were unable to demonstrate any relationship between AMF colonization or richness and the cover or richness of plant species from typically mycorrhizal or nonmycorrhizal families. Classification of plant families on the tendency of members to form mycorrhizas (Gerdemann, 1968; Newman & Reddell, 1987) may be useful for general discussions on relationships between plants and mycorrhizal fungi; however, these classifications are not absolute. Some members of typically nonmycorrhizal plant families can form mycorrhizae at some point in their life history or under certain environmental conditions (Muthukumar et al., 2004; Plenchette & Duponnois, 2005). Moreover, some plants from typically mycorrhizal plant families, such as many of the annual species encountered in this study, may not benefit from associating with mycorrhizal fungi because of their short life-span and ruderal nature (Collier et al., 2003). In this semiarid riparian setting, where annual species are a major component of the flora, the life history characteristics of the above-ground vegetation appear to be a better predictor of AMF richness than the membership of species of typically mycorrhizal or nonmycorrhizal families.

Findings from this study indicate that environmental gradients also influenced species composition of AMF communities found in riparian Populus–Salix stands. Many AMF species appear to have wide ranges of environmental tolerance (Koske, 1987; Stahl & Christensen, 1991), but the presence of some species appears to be restricted by environmental variables. Ordination identified several variables related to floodplain topography, including plot elevation above and distance from the active channel and stand age, which may have some influence on the composition of the AMF community. Some species showed a clear affinity for young, moist fluvial surfaces while others were more likely to occur in drier sites higher on the floodplain. Most AMF species identified as riparian affiliates (60% or more of their occurrences in wet sites) have not been detected in other studies of the surrounding Sonoran desert (Stutz & Morton, 1996; Stutz et al., 2000). These results are similar to a study of South Carolina wetlands where different AMF dominated the dry, intermediate and wet portions of the moisture gradient (Miller & Bever, 1999).

Some studies of AMF in relation to environmental gradients have found decreasing AMF activity and richness with increasing soil phosphorus concentration (Tanner & Clayton, 1985; Johnson et al., 1991; Amijee et al., 1993; Wetzel & van der Valk, 1996; Kennedy et al., 2002), and another found that AMF richness was positively correlated with soil nitrogen concentration (Landis et al., 2004). Our study identified exchangeable potassium as having a relationship to AMF community composition but did not find a relationship between soil phosphorus or nitrogen concentration and AMF species richness, degree of colonization, or community composition.


The AMF community composition of Populus–Salix stands along the Verde River is similar to that of the surrounding desert, but also contains a subset of unique species. Several interrelated successional, positional, edaphic and vegetation variables were important in structuring the AMF community. AMF richness decreased with increasing Populus–Salix stand age and plot elevation above and distance from the active channel. AMF richness and degree of colonization were highest in young Populus–Salix stands occupying low, moist areas on the floodplain. AMF richness showed no relationship with cover and richness of all herbaceous species considered together, but was positively related to the cover and richness of perennial plant species. Some AMF species showed a clear affinity for young, moist fluvial surfaces and many of these ‘riparian affiliate’ species have not been encountered in previous extensive studies of the surrounding Sonoran Desert.


Site access for this project was provided by the Prescott and Tonto National Forests, Phelps Dodge, Dead Horse Ranch State Park and the Fort McDowell Yavapai-Apache Reservation. This work was funded by an EPA STAR Fellowship (U915369) and a Society for Wetland Scientists Student Research Grant to VBB. Jim Bever and two anonymous reviewers contributed valuable comments on earlier versions of this manuscript.