Prosopis juliflora is part of the Prosopis juliflora-pallida complex, a taxonomically complicated, interrelated, and controversial group both in the native and non-native ranges of species in this complex [5]. Because of this pervasive taxonomic confusion, we took a conservative approach to sampling in Venezuela, native range of P. juliflora, and in Peru, the native range of what is most likely to be Prosopis pallida [5]. Our intent was to make a broader measurement of the impacts of the Prosopis juliflora complex in the native ranges by using both ends of the species that make up the P. juliflora-pallida complex. We located two field sites in Peru, (i) Chulucanas (S 05°11′ 49.4″; W 080°16′ 53.3″; 174 m) and (ii) Catacaos (S 05°27′ 06.5″; W 80°14′ 37.2″; 253 m). We located two field sites in Venezuela, in the native range of Prosopis juliflora [5], (i) Mucumi (N 08°30′ 10.0″; W 71°21′ 53.0″; 1000 m) and (ii) Puente Real (N 08°28′ 41.0″; W 71°24′ 38.0″; 650 m). Both sites are within the Lagunillas Semi-arid Enclave, which is an inter-Andean enclave, the largest in Venezuela with an area of 262 km². The vegetation at both study sites is classified as thorn scrub. In these sites the dominant elements in the upper stratum were the thorny shrubs Prosopis juliflora and Acacia macracantha, which form an open and discontinuous canopy up to 3 m in height. The columnar cacti Stenocereus griseus, Pilosocereus tillianus and Cereus repandus are emergent elements (up to 5–7 m in height), which tend to be spatially associated with these leguminous shrubs. The intermediate stratum (up to 50 cm) is formed by shrubs of Croton spp., Capparis odoratissima, herb Hibiscus phoeniceus, cacti Opuntia caribea, Opuntia depauperata, Opuntia aff. elatior and Solanum spp. A bromeliacea of the Pitcairnia genus, the herbs Jatropha gossypifolia, Cnidosculus urens and Evolvulus sp., and annual plant species were also present. Three field sites were located in India and two sites were located in Honolulu, Hawaii, non-native ranges of P. juliflora. The sites located in India were: (i) Delhi (N 28°35′ 58.6″; E 077°10′ 15.6″; 233 m), (ii) Kutch, Gujarat (N 23°55′ 58.2″; E 069°48′ 50.4″; 407 m), and (iii) Jaipur, Rajasthan (N 27°01′ 00.3″; E 075°46′ 12.9″; 484 m). The three Indian sites were chosen to represent a range of climates, with Delhi, Gujarat and Rajasthan having humid subtropical, tropical monsoon and tropical arid climates, respectively. The two sites in Hawaii were: (i) Sand Island (N 21° 18′ 12.5″ W 157° 52′ 51.7″ and (ii) Iroquois Point-Hau Bush (N 21° 19′ 24.9″; W 157° 58′ 16.7″ and N 21° 18′ 20.9″; W 157° 1′ 52.5″), both at sea level.

At each site in Peru, we randomly placed 1 m² quadrats in areas of large patches of P. pallida and where P. pallida was not present, what we refer to as ‘open areas’. We recorded the number of species in each plot. In 2008, we sampled 96 pairs of plots at Chulucanas and 92 pairs at Catacaos. In 2009, we sampled 72 pairs in Chulucanas and 120 pairs at Catacaos. At each site in Venezuela in July 2012, we randomly placed one 1 m² quadrat under each of 15 different P. juliflora trees and in open grassland adjacent to each tree where P. juliflora was not present and recorded the number of species in each plot. At the Delhi site P. juliflora primarily occurs in dense stands (3.6 trees per 25 m²). Here in May 2010 we established 10 randomly placed 5×5 m² quadrats under P. juliflora, with each plot representing the understory of a different set of several trees, and 10 5×5 m² quadrats in open area outside of the P. juliflora stands. At Gujarat, the site was a scrub forest of P. juliflora (1.4 trees per 25 m²) in which 7 randomly placed 5×5 m² plots were located under P. juliflora and 7 in open areas away from P. juliflora canopy, during June 2008. At Rajasthan, P. juliflora tended to be smaller and occur in very large closed canopy patches precluding pairing local sub canopies and open patches (4.4 trees per 25 m²). Thus 10 randomly placed 5×5 m² quadrats were located under P. juliflora roughly in the centre of a typical stand that was several kilometres in diameter, with control quadrats (no P. juliflora) approximately 2 km away, from the quadrats with P. juliflora but in what appeared to be very similar environmental conditions. Sampling was in June 2010. At Delhi and Rajasthan, open areas were selected outside P. juliflora stands because these stands were closed-canopy woodlands rather than savanna-like vegetation. At Delhi, the open area was characterized by few far-separated trees of Albizzia lebbeck, Balanites roxburghii with interspersed patches of shrub species including Grewia sp., Capparis sepiaria, Capparis decidua and Zizyphus nummularia, and grasses including Cenchrus ciliaris, Chrysopogon fulvus and Heteropogon sp. At Rajasthan also, open area consisted of far-separated tree species (Dalbergia sissoo and Azadirachta indica ) with interspersed patches of shrubs (Zizyphus sp., Lantana camara, Morus sp.) and grasses (Saccharum sp. and Cenchrus ciliaris). However at Gujarat, open areas and P. juliflora subcanopy quadrats were interspersed within the P. juliflora stand due to non-overlapping canopies and open areas included tree species (Acacia senegal and Wrightia tinctoria), shrub species (Euphorbia caducifolia, Commiphora wightii and Grewia sp.) and herb species (Tephrosia sp., Rhynchosia sp. and Cenchrus sp.).

Prosopis juliflora was first recorded at the Sand Island site in Hawaii in the early 1970’s and today this area holds the largest population of Prosopis juliflora and P. juliflora x P. pallida hybrids in Hawaii. The second site, Iroquois Point-Hau Bush, is within a small stretch of undeveloped coastal area. Its location is approximately 10 km to the west of the Sand Island site with a very similar climate to that of Sand Island. Within this area, there is a large established population of P. pallida with P. juliflora and hybrids found near the coast. In October-November 2011, 5×5 m² quadrats were placed at each site in areas of dense infestation of P. juliflora and areas where it was not present. The number of plant species present in each quadrat was counted.

Our initial plan was to use 1 m² plots in the two non-native ranges, India and Hawaii, as we did in the native range, of P. juliflora, Venezuela and of P. pallida, Peru. However, the number of species present was much lower under canopies of P. juliflora in the non-native ranges, India and Hawaii. We therefore used larger plots to avoid large numbers of zeros in these plots. We did intra-site species richness comparisons under P. juliflora canopies and nearby open areas where we used identical sampling methods, at each site in native and invaded range and then compared the effect of P. juliflora on species richness observed in its invaded ranges to those observed in its native range.

Species richness in the open and under P. juliflora canopies at each site was compared using independent samples t-tests, one-way ANOVA or Mann-Whitney U tests [29].

No specific permits were required for the field studies carried out in Venezuela, India or Hawaii, USA; and field studies in these areas did not involve endangered or protected species. For the field studies carried out in Peru, permission was obtained from the Agriculture Ministry – Perú (N°145-2008-INRENA-IFFS-DCB). The field study did not involve endangered or protected species.

We compared species richness beneath P. juliflora canopies to that beneath canopies of the native congener, P. cineraria, at Deer Park at Patiala, Punjab, India (N 30°17′ 00.2″; E 076°23′ 07.5″; 239 m). The area has a sub-tropical climate and total annual rainfall during the last 5 years of 703.6 mm. The selected site has been invaded by P. juliflora that reach heights of 12 to 14 m and densities of 1.1 trees per 25 m². Prosopis cineraria occurs there as scattered individuals reaching heights of 4 to 9 m and densities of less than 0.4 trees per 25 m². The canopy areas occupied by individual P. juliflora and P. cineraria trees were 139.9 and 20.8 m², respectively. We used slightly smaller quadrats (4×4 m²) for measurements at this site so that a single quadrat represented a single tree either of P. juliflora or P. cineraria. In October 2010, we randomly selected 10 trees of P. juliflora or P. cineraria of similar sizes and laid a 16 m² quadrat under the canopy of each tree keeping the trunk in the centre, and recorded the number of plant species in each quadrat. We also located 10 randomly located 16 m² quadrats in open areas where neither P. juliflora nor P. cineraria was present. Differences in species richness in open, under P. juliflora and P. cineraria were tested using one-way ANOVA and post-hoc Tukey’s test [30].

We also compared soils from the rhizospheres of P. juliflora and P. cineraria at Deer Park. We did sampling at this site during two time periods, November 2009 and March 2011. During November 2009, soil was collected from rhizosphere of 4 individual trees of each Prosopis species and from 4 nearby open locations (n = 4 for each of the three treatments). Soil was collected near tree trunk to a depth of 30 cm and presence of a mesh of horizontal and vertical roots in the pit confirmed that it was rhizosphere soil. In March 2011, soil was collected from rhizosphere of 6 individual trees of each Prosopis species and from 6 nearby open locations (n = 6 for each of the three treatments). Soil was collected from 3 points around the tree trunk of each tree selected (no. of subsamples per tree = 3). The soil was then air-dried, sieved (2 mm), and stored in paper bags for later analyses. Five g soil from each sample was shaken with 25 mL water for 1 h followed by filtration through Whatman # 1 filter papers. pH and electrical conductivity were measured in soil extracts using a digital meter (Metrex Sci. Instr. Pvt. Ltd., India). Organic carbon was estimated following methods of Walkley and Black [31]. Total organic nitrogen was estimated using semi-micro Kjeldahl digestion [32]. We estimated PO₄^3–P in 2.5% acetic acid soil extracts through molybdenum blue method [32]. Total phenolic content of soils was measured with the Folin-Ciocalteu method [33].

Data from collection during two time periods were combined (n = 10 for each treatment) and, tested for the differences among open, P. juliflora and P. cineraria soils for each soil chemical property using one-way ANOVA and post-hoc Tukey’s tests [29].

We first tested the effects of leaf litter at The University of Montana in a greenhouse experiment. We compared the effects of litter of P. juliflora and P. cineraria on the mortality of six native Indian species (Acacia nilotica, Brassica campestris, Brassica juncea, Chloris dolichostachya, Dalbergia sissoo and Prosopis cineraria). Litter from the Prosopis congeners were collected from three different trees in Aravalli Biodiversity Park, Delhi, India (N 28° 33′ 18.9″; E 077° 08′ 56.8″; 247 m), mixed, and then air dried. Rocket pots (200 cm³, 3 cm diam) were filled with 20/30 grit silica sand, and 5–10 seeds were planted 5 mm below the surface of the sand in each pot. We put 1.0 gm of litter from either P. juliflora or P. cineraria on the surface of the substrate, so that the seeds would have to germinate and grow through the leaves as they would through litter in the field. As seeds germinated they were thinned continually so as to keep only the largest seedling in each pot. The experiment started on 8 January 2010 and ended on 9 February 2010. We summed the total final mortality for all of the remaining largest seedlings in all pots for each species, and then compared treatments using the means of mortality for the six species as replicates in an ANOVA testing treatments (control, P. juliflora, P. cineraria) and followed by Tukey HSD tests [29].

We also tested the effects of soil amended with leachate from leaves of the two congeners on three crop species, B. campestris, B. juncea and Sorghum bicolor, at the University of Delhi. We used these three crop species because these are common crop species in north-west India where P. juliflora has invaded agricultural fields or is present at the boundaries. Prosopis cineraria is often present in crop fields (Figure 1F). Leaves of the two Prosopis species were collected from the same sites as in the previous experiment and air dried prior to use. Ten g of air dried leaves of each were soaked separately in 100 mL distilled water for 14 h in the dark followed by filtration through Whatman # 1 filter paper. Soil (sandy loam) for the bioassay was collected from an area not occupied by P. juliflora, and at the same site where leaves were collected. 50g air-dried soil was weighed into a 9 cm Petri dish and treated with 15 mL distilled water, 5x, 2x, or 1x diluted P. juliflora or P. cineraria leaf leachate which yielded concentrations of leaf leachate in soil at 0, 60, 150 or 300 µL of leaf leachate/g soil. Brassica campestris and B. juncea seeds were washed with distilled water before use. Sorghum bicolor seeds were surface-sterilized by soaking in 1% sodium hypochlorite for 10 min and then rinsing in distilled water 20 times before use. For each of the seven treatments and three test species, we used six Petri dishes. Ten seeds of B. campestris, B. juncea, or S. bicolor were placed in each Petri dish and kept at 20–23°C. Seven days after treatment application we measured the shoot and root lengths of the germinants. Two-way ANOVA was used to test the effect of leachate source and concentration as fixed factors on root length. We first compared the effect of P. juliflora leachate or P. cineraria leachate treatment (each concentration) with control and then the effect of P. cineraria leachate with P. juliflora leachate applied at identical concentrations, using independent samples t-tests [30].We also correlated leachate concentration with root length for each bioassay species [30].

The biochemical effects of litter from different species or different regions could be confounded by different effects of the pathogens in the litter. Therefore we cultivated bacteria and fungi from the leaves of the two Prosopis congeners and applied them to two species native to India, Brassica juncea and Chloris dolichostachya. Cultivation of bacterial and fungal communities of Prosopis leaves was achieved by streaking the leaves of each congener (in triplicate) across the surface of several standard microbiological media agar plates. Plates were incubated at 37°C for 1 week in order to observe any potential bacterial or fungal growth. Five different media were used to cultivate microbial growth. The media used to select for bacteria were R2 agar [34], Minimal Agar, and Nutrient Agar [35] treated with 300 mg L⁻¹of chloramphenicol to inhibit fungal growth. The media used to select for fungi were Mycobiotic Agar and Potato Dextrose Agar [35]. After the incubation period, individual colonies were selected from the media plates and were used to make separate bacterial and fungal community inocula from each Prosopis congener. Each inoculum (either bacterial or fungal) was prepared by aseptically transferring individual microbial colonies into sterile Eppendorf tubes (2 mL volume) filled with 1.5 mL sterile Milli-Q water, and gently vortexed. Mixing 0.5 mL of the bacterial community inoculum with 0.5 mL of the fungal community inoculum made a third composite microbial inoculum. Individual seedlings of B. juncea and C. dolichostachya were germinated in 200 cm³ rocket pots filled with 20/30 grit silica sand, and after two weeks were inoculated with the bacterial, fungal, or composite microbial inocula, with n = 15 for each species and treatment combination. We measured survival of these seedlings 20 days after inoculation.

To determine the effect of leaf leachate treatment on soil chemistry, 50g soil in a 9 cm Petri dish was amended with 15 mL distilled water (control) and 1x diluted P. juliflora or P. cineraria leaf leachate (0 and 300 µL P. juliflora or P. cineraria leaf leachate/g soil) as described above. Six replicate Petri dishes were used per treatment. Immediately after treatment, lids of Petri dishes were removed and soil was allowed to air-dry for 48 h. Soil was then analysed for pH, electrical conductivity, organic carbon, PO₄^3–P, and total organic N as described above. We tested the difference among P. juliflora, P. cineraria leaf leachate amended and unamended soil for each soil variable using one-way ANOVA and post-hoc Tukey’s test [30].

Soil amended with leaf leachates of P. juliflora or P. cineraria, as described above, was also analyzed for total phenolic concentration in addition to other chemical properties. In a separate experiment, we added 60 mg of air dried leaves of P. juliflora or P. cineraria to 5 g soil and moistened the soil with 2 ml distilled water. Litter-amended soil was incubated at 30–34°C for 0, 1, 2, 3, 4, 6, 8, 10 or 14 d. Each treatment and the control (unamended soil) were replicated four times. Measurement of total phenolics in soils from two experiments was performed as described above. We tested the differences among P. juliflora litter, P. cineraria litter, leaf leachate amendments, and untreated soil for total phenolics using one-way ANOVA and post-hoc Tukey’s test [30].

Nakano et al. [27] reported that the allelopathic effects of P. juliflora foliage appeared to be caused by L-tryptophan, thus we suspected it might play a role in the effects we measured for litter and leaf leachates. For this, we quantified L-tryptophan in leaf leachates from P. juliflora or P. cineraria and in soil amended with these leachates. Two ml of leaf leachate from P. juliflora or P. cineraria was added to 10 mL methanol, and L-tryptophan was quantified with high-performance liquid chromatography (HPLC). Five g soil was treated with 2 mL leaf leachate from P. juliflora, P. cineraria or with distilled water and then immediately extracted in 10 mL methanol by shaking for 20 min followed by centrifugation at 4500 rpm for 10 min. The supernatant was filtered through 0.2 µm PES syringe filters directly into HPLC vials.

L-tryptophan level in methanol extract of each soil and leachate sample was determined by using HPLC (Waters Corp., Milford, U.S.A), employing a Waters Spherisorb 5 µm ODS2 column (4.6×250 mm Analytical Column). The mobile phase consisted of 50 mM sodiumphosphate buffer, pH 3.5 containing 20% methanol (v/v) and flow rate was 1 mL/min. L-tryptophan was detected at 214 nm using a photodiode array detector. 10–20 µL of sample was injected in the partial loop needle overfill mode, run time for each sample was 20 min, and the retention time of L-tryptophan was 8.17 min with 0.27% RSD. L-tryptophan concentration in leaf leachates from two congeners was compared using independent samples t-test [30].

Plant species richness in the understory of P. juliflora in its native range of Venezuela was 48% and 63% higher than in nearby open areas, at Puente Real and Mucumi, respectively (Figure 2, middle panel; F_{understory vs. open} = 324.0; df = 1,56; p = 0.035). Plant species richness in the understory of P. pallida in its native range of Peru was 27% and 44% higher than in nearby open areas, at Chulucanas and Catacaos, respectively in 2008 (Figure 2, upper panel; t _Chulucanas = 5.13; df = 1,94; p<0.001; t _Catacaos =  = 2.67; df = 1,92; p = 0.011). However, in 2009 richness was 28% and 25% lower than in the open at those same sites (Figure 2, upper panel; t _Chulucanas = 2.70; df = 1,70; p = 0.009; t _Catacaos =  = 4.678, df = 1, 118, p<0.0001). Analyzed over both years P. pallida had no effect on understory richness in its native range in Peru (F_canopy = 0.79; df = 1,374; P = 0.778).

In the non-native range of India, species richness in P. juliflora invaded areas was 32%, 54% and 78% lower at Delhi, Gujarat, and Rajasthan, respectively, compared to areas without P. juliflora (Figure 2, lower panel: t_Delhi = 2.72; df = 1,18, p = 0.014; t_Gujarat = 3.37; df = 1,12; p = 0.006, and for Rajasthan, Mann-Whitney U test, p = 0.012). The mean effect in India over all three sites was a decrease in species richness of 56%. In habitats invaded by P. juliflora in India native species accounted for 92%, 67% and 100% of the total number of species present below P. juliflora canopies at sites Delhi, Rajasthan and Gujarat, respectively. In Honolulu, Hawaii species richness in P. juliflora invaded areas was 70% and 65.3% lower at Sand Island and Iroquis point, respectively (Figure 2, lower panel: t_{Sand Island} = 5.93, df = 1,18, p<0.001; t_{Iroquis point} = 7.49; df = 1,18, p<0.001). In Hawaii, however, more than 90% of understory species were non-native.

At a fourth site in India, species richness did not differ beneath canopies of P. cineraria relative to open areas, but species richness was 63% lower under P. juliflora canopies than in the open and 50% lower than under P. cineraria canopies (Figure 3; F = 10.018; df = 2,27; P<0.001).

The pH of P. cineraria soil was significantly lower than that of soil in the open (Figure 4A: F_pH = 15.914; df = 2,27; p<0.0001, Tukey, p<0.0001) and P. juliflora soil (Tukey, p = 0.001) but that of P. juliflora soil did not differ from soil in open (Tukey, p = 0.525). As compared to soil in open, organic carbon and nitrogen were significantly higher in P. juliflora (Figure 4C; F_{Organic carbon} = 9.798; df = 2,27; p = 0.001, Tukey, p = 0.001; Figure 4E: F_Nitrogen = 19.897; df = 2,27; p<0.0001, Tukey, p = 0.001) and P. cineraria soils (organic carbon: Tukey, p = 0.005; nitrogen: Tukey’s test, p<0.0001) but did not differ significantly between the two congeners (organic carbon: Tukey, p = 0.775; nitrogen: Tukey, p = 0.109). Electrical conductivity, exchangeable PO₄^3–P and total phenolic content were significantly higher in soil from under P. juliflora than soil from under P. cineraria (Figure 4B: F_{Electrical conductivity} = 9.105; df = 2,27, p = 0.001, Tukey, p = 0.006; Figure 4D: F_Phosphorus = 26.574; df = 2,27, p<0.0001, Tukey, p<0.0001: Figure 4F: F_Phenolics = 13.711; df = 2,25, p<0.0001, Tukey, p = 0.020) or from the open (electrical conductivity: Tukey, p = 0.001; exchangeable PO₄^3–P: Tukey, p<0.0001; total phenolic content: Tukey, p<0.0001). These factors did not differ significantly between P. cineraria soil and soil from the open (electrical conductivity: Tukey, p = 0.847; exchangeable PO₄^3–P: Tukey, p = 1.000; total phenolic content: Tukey, p = 0.084).

Leaf litter from the Indian native P. cineraria increased the mortality of Dalbergia sissoo, but had no effect on any other native Indian species (Figure 5). In contrast, leaf litter from the invasive P. juliflora increased the mortality of all six native Indian species from 14–44%. The mean mortality of Indian species when exposed to litter from P. cineraria (5±5%) was not significantly different than that in the control (ANOVA, F_treatment = 15.09; df = 2,15; P<0.001; Tukey P = 0.970), but much lower than the mean mortality (33±4%) of plants treated with P. juliflora litter (Tukey P = 0.001).

As compared to control, leachate from the leaves of the native P. cineraria significantly enhanced the root growth of B. campestris at 150 (Figure 6 upper panel; t_{0 vs. 150} = −2.561; df = 1,10; p = 0.028;) and 300 µL/g soil concentrations (t_{0 vs. 300} = −2.955; df = 1,10; p = 0.014), and of B. juncea at 60 (Figure 6 middle panel; t_{0 vs. 60} = −3.506; df = 1,10; p = 0.006), 150 (t_{0 vs. 150} = −4.759, df = 1,6; p = 0.003), and 300 µL/g soil concentrations (t_{0 vs. 300} = −3.067; df = 1,6; p = 0.025). The root length of S. bicolor seedlings grown in P. cineraria leaf leachate amended soil was not significantly different from control at any tested concentration (Figure 6 lower panel; t_{0 vs. 60} = 1.475; df = 1,10; p = 0.171; t_{0 vs. 150} = −1.370, df = 1,10; p = 0.201; t_{0 vs. 300} = 0.496; df = 1,10; p = 0.631). In contrast, leachate from the leaves of the invasive P. juliflora significantly reduced the root growth of B. campestris as compared to control at the highest concentration (Figure 6 upper panel, t_{0 vs. 300} = 3.893, df = 1,10, p = 0.003) and of S. bicolor at 150 (Figure 6 lower panel, t_{0 vs. 150} = 3.006, df = 1,10, p = 0.013) and 300 µL/g soil concentrations (t_{0 vs. 300} = 6.226, df = 1,10, p<0.0001). Brassica juncea root growth increased in soil treated with 60 (Figure 5 middle panel; t_{0 vs. 60} = −2.477, df = 1,10; p = 0.033) and 150 µL/g soil concentrations (t_{0 vs. 150} = −3.693, df = 1,10, p = 0.004) of P. juliflora leachate and was not affected at highest concentration (t_{0 vs. 300} = 0.002, df = 1,10, p = 0.999).

Root length of seedlings of all 3 species grown in soil amended with leaf leachate from P. juliflora was significantly lower than that of seedlings grown in P. cineraria leaf leachate-amended soil at similar concentrations i.e. 150 (Figure 6 upper panel; t_{B. campestris =}2.764; df = 1,10; p = 0.020; t_{S. bicolor =}3.116; df = 1,10, p = 0.011) and 300 µL/g soil (t_{B. campestris =}5.573; df = 1,10; p<0.001; t_{B. juncea =}2.936; df = 1,10; p = 0.015; t_{S. bicolor =}4.963; df = 1,10; p = 0.001), with the exception of B. juncea at 150 µL/g soil (t_{B. juncea =}1.151; df = 1,10; p = 0.277). However at 60 µL/g soil concentration, there was no significant difference in root growth of all 3 species grown in P. juliflora leaf leachate-amended soil compared to those in P. cineraria leaf leachate-amended soil.

There was a significant interaction effect between leachate source and concentration on root growth for all 3 species (Figure 6). We also found significantly negative correlations between P. juliflora leachate concentration and root growth of B. campestris (r = −0.681, p<0.001) and S. bicolor (r = −0.476, p = 0.019), however no significant correlation was observed in case of B. juncea (r = −0.060, p = 0.779). In contrast, significant positive correlations between P. cineraria leachate concentration and root growth of B. campestris (r = 0.594, p = 0.002) and B. juncea (r₌0.589, p = 0.002) were observed, however no significant correlation was observed in case of S. bicolor (r = 0.084, p = 0.695).

Bacteria and fungi isolated from the leaves of P. julifora and P. cineraria, when applied directly to other species, killed almost all B. juncea and C. dolichostachya seedlings precluding statistical analysis, but there were no discernible patterns in mortality for isolates from the two congeners. Fungal isolates from P. juliflora, applied as a group, killed 97% of B. juncea and 98% of C. dolichostachya seedlings (data not shown). Bacterial isolates from P. juliflora, applied as a group, killed 100% of B. juncea and 94% of C. dolichostachya. Fungal isolates from P. cineraria, applied as a group, killed 98% of B. juncea and 100% of C. dolichostachya (data not shown). Bacterial isolates from P. cineraria, applied as a group, killed 94% of B. juncea and 99% of C. dolichostachya.

Both treatments, P. juliflora leaf leachate and P. cineraria leaf leachate, resulted in a significant decrease in pH and an increase in EC, OC, PO₄^3–P, and total organic N of soil with the effect of former treatment being significantly higher than latter (Table S1). Prosopis cineraria leaf leachate treated soil did not differ significantly for PO₄^3–P from untreated soil (Table S1).

Soil amended with leaf leachate of P. juliflora had 12.4 times higher values of total phenolics (5.956±0.054 mg/100 g soil) than soil amended with P. cineraria (0.479±0.019 mg/100 g soil) and 60.2 times higher concentrations than unamended soil (0.099±0.004 mg/100 g soil; F = 9734.99; df = 2,15; p<0.015; Figure 7A). The total phenolic content of soil amended with leaf litter from P. juliflora was significantly higher than soil amended with P. cineraria litter and unamended soil, from 0 d to 14 d of incubation (Figure 7B, Table S2).

L-tryptophan concentrations in the leachates of leaves of P. juliflora (4.92±0.05 µg/ml leaf leachate) was 73.1% greater than that of P. cineraria (1.32±0.014 µg/ml leaf leachate) (t = 70.509, df = 4, p<0.0001). Over 40% (0.79±0.006 µg/g soil) of the originally added 1.968 µg/g soil of L-tryptophan from P. juliflora leachate, was recovered from soil immediately after application, but we could not detect L-tryptophan in soil amended with P. cineraria leachates.