Assessing Water Use Efficiency, Soil Health, and Pollinators within a Transition from Irrigation to Dryland Management in the Texas High Plains

A. SARE Report on Soil Health

The decline in groundwater supply in the Texas High Plains due to historical excessive irrigation within the past 20 years, but more frequently due to extreme droughts, is forcing some growers to convert center-pivot irrigated cropland to dryland production.

This SARE funded research project assessed the changes in soil health indicators (chemical and biological indicators) during 5 years in fields representing two transition scenarios: (a) from high irrigation method to low irrigation method (center pivot to subsurface drip) and (b) from high irrigation method to dryland (center pivot to dryland), in comparison to continuous center-pivot management.

Our study began in October 2018, and we continued to take soil samples every year until 2022. The study was conducted in Lamb County near Littlefield, TX (approximately 34.00 N, 102.36 W), in fields operated by three different growers (Table 1). The maximum distance between fields was 45 km. Irrigation is required in this region to meet crop water requirements owing to low average annual precipitation of about 470 mm (Lascano, 2000). These growers participated in the Environmental Quality Incentives Program (EQIP), a USDA-NRCS cost-share program, to convert some of their center-pivot (CP) irrigation area to SDI, while retaining other areas under CP and converting some to dryland production.

There are two phases of our study. Our first publication, which acknowledges this SARE grant (LS 20-341), reported the changes in soil health within the first 3 years (short-term) changes of these transitions.  The publication can be cited as: Bhandari, K. B, Acosta-Martínez, V., Pérez-Guzmán, L., & West, C. P.  (2022). Soil health within transitions from irrigation to limited irrigation and dryland management. Agricultural & Environmental Letters, 7, e20077. https://doi.org/10.1002/ael2.20077

During the first-three year’ transitions, there were three main findings and points elucidated from our soil health assessment:

• We found declines in soil water content, potassium (K), sodium (Na), and soil organic carbon with transition from irrigation to reduced irrigation and dryland.
• Severe drought in the final year revealed reduced amounts of multi-enzyme activities, total ester-linked fatty acid methyl ester (EL-FAME), and total fungi.
• Transitioning to low water-input management in this environment complicates efforts to maintain microbial components of soil health. Longer-term comparisons are needed to detect slow changes in soil health indicators on producers’ fields.'

After our first paper reporting on the first three years of our study, our samplings continued to evaluate soil health as affected by long-term transitions (5 years) to dryland in the Southern High Plains. Over time, some of the producers had to change their management for different reasons especially not use center pivot irrigation due to the below average precipitation received and extreme heat years in this region within the past 10 years. We still evaluated management transitions from center pivot to drip irrigation or center pivot to dryland as shown in the table below.  These fields and irrigation scenarios were used in pollinator sampling to assess native bee communities across fields (Table 2).

Our soil health assessment continued to use total FAMEs as indicators of soil microbial community size, microbial community composition based on bacterial, fungal and AMF (arbuscular mycorrhizal fungi) markers, and enzyme activities involved in carbon, nitrogen, phosphorus, and sulfur cycling from the CNPS assay.

Our overall assessment for the first transition scenario on the conversion of center pivot to drip irrigation as shown in the figure below demonstrated no significant change in the total FAMEs or the microbial community size over all 5 years.  We also did not find any major change in soil bacteria and fungi, including AMF, a type of fungi that when associated to plant roots, will increase their nutrient and water uptake from soil. The trends were similar for CNPS activity representing several enzyme activities involved in nutrient cycling key for plant nutrition. These trends could mean that the transitions from center pivot to drip irrigation in this region and/or type of soils may not represent significant changes in water available in soils and provided to soil microbes and their enzymatic capacity.

Our overall assessment for the second transition scenario on the conversion of center pivot to dryland evaluated (Fig. 1), demonstrated no significant (consistent) change in the total FAMEs or microbial community size over all 5 years.  We also did not find any major change in soil bacteria and fungi, including AMF, a type of fungi that when associated to plant roots, will increase their nutrient and water uptake from soil. The trends were similar for CNPS activity representing several enzyme activities involved in nutrient cycling key for plant nutrition. In this transition, where we know that less water is being delivered to the soils as they are transitioning to dryland, we acknowledge the need to select management practices that could maintain these communities and these key enzymatic processes in the future.

Figure 1. Total FAME, Bacterial Sum, AMF, Fungal Sum and CNPS enzyme activity measured across agricultural fields undergoing transition from center pivot irrigation to drip irrigation (left) and dryland (right).

Overall findings of the soil health study:

• Our 5 year-study provided a soil health assessment framework to evaluate soil health in the fragile soils of the Southern High Plains that continue to be used as several fields transition to dryland. Our soil health assessment is based on methodology that can establish the linkage of soil microbial communities and key processes related to soil organic matter dynamics and biogeochemical cycling because they can influence soil conservation.
• Within our study, the region experienced a drought from 2018 to 2020, and our soil health assessment demonstrated decreases in the soil microbial communities due to this additional layer of factors that can impact soil microbial communities and the soil health. It is expected an intensification of more severe and extreme climatic conditions in this region, and the soil health assessment framework established in our study will be key to better address the effects of climate extremes in overall soil health and essential functions that microbes provide. As the studies continue in other sites, we will be able to evaluate management practices that could buffer and minimize the detrimental effects of climate extremes.
• Our study is delivering key data on soil microbial components that will key to be available for future comparisons on the transition to dryland

B. SARE Report on Native Bee Communities: Biodiversity and Relationships with Irrigation Scenarios and Soil Conditions

Despite the likely importance of pollinators in cotton, corn, and sorghum production, an understanding of pollinators occurring in these crops in the Texas High Plains and their contributions to the production of these crops remain incomplete. Furthermore, in this intensively managed agricultural region, wild bees and pollination services are vulnerable to non-target agrochemical exposure, which can in turn impact crop production and native biodiversity. Thus, understanding pollinator communities across a transition related to irrigation practices is an important component in selecting management practices in field and forage crops. Importantly, understanding how different irrigation practices directly influence soil habitats and resident nesting and developing bees is a vastly understudied area of pollinator research. The Southern High Plains provides both the bee biodiversity, and the cropping transition types necessary to investigate these processes.

The pollinator focused objectives of the work included to 1) document biodiversity across a limited subset of farms in the Southern High Plains and 2) determine differences in bee communities across designated irrigation scenarios involving shifts from dominant center pivot irrigation to subsurface drip and dryland production fields.

Sampling was conducted on five dates (June through September) from 2021-2022 across three different farms (Designated Dg, Mu, and Ma, Table 2 above). The number of fields across farms varied, with the goal to replicate fields having one of the three types of irrigation scenarios (i.e. center pivot, subsurface drip, and dryland). In addition to these fields, three additional transects were included because of the dominance of coarse woody debri occurring at three edge-of-field locations on the partner farms. For this sampling, the general hypothesis was that these areas could provide refuge and resources for wild bees (e.g., non-crop flowering vegetation, soil nesting sites, and lower soil surface temperatures).

At each field, 60 m linear transects were marked and bee bowls (i.e., 3.5-ounce plastic cups) were placed at 5 m intervals along the transect. Bee bowl contents were retrieved after 24 hours and specimens were processed by pinning or by preserving in 70% ethanol in glass vials. Along each transect and opportunistically at field edges, the soil surface and any bee activity were observed in order to locate potential ground nesting bee habitats.

In the lab, native bee morphospecies were analyzed, documented, and tallied according to each collecting date and field. For analysis, bee counts were combined across dates to yield a native bee community sample for each field and according to the different irrigation-transition scenario (Table 2).

An initial summary of the total diversity of native bees was developed across all fields. Bee morphospecies were ranked according to total abundances and total relative abundances. A site by species data matrix (12 sites by 71 morphospecies) was used to calculate 1) S; the total number of non-zero cells and morphospecies for each row/field, 2) E; Evenness of the native bee community, 3) H; Shannon index, and 4) D; Simpsons diversity index. These measures were used to evaluate relationships with soil health variables using principal components analysis. The site by species matrix was used to further conduct an analysis of the native bee communities occurring across fields and irrigation scenarios using non-metric multidimensional scaling (NMDS). This method is robust to community datasets containing an inflated number of empty cells (i.e., with values of zero). Prior to NMDS analysis and to remove noise in the community data, native bees occurring with less than 3 non-zero values were removed from the dataset prior to analysis, which resulted in a reduction of native bee morphospecies from 71 to 29. Using NMDS, a second matrix containing soil variables was analyzed to determine if soil variables were related to potential bee community groupings. To aid interpretation, sites were labelled according to both the field/farm number and with the current irrigation scenario (using symbols). Finally, indicator species analysis was used to determine native bee morphospecies affinities for irrigation transition scenarios. 

Results

Native Bee Communities

Across all farms and fields, a total of 4427 individual bees were collected across four bee families: Halictidae, Andrenidae, Apidae and Megachilidae (Table 3). The family Apidae contained the greatest number of morphospecies, followed by the Andrenidae. A total of 31 genera were collected across all families and 71 morphospecies were reported. Of these 71 morphospecies, 38 were confirmed at the species level. The genus Perdita contained 17 individual morphospecies, these were sorted by careful examination of the facial and tergal patterns (see Fig. 11). The genera Triepeolus and Lasioglossum contained five different morphospecies. Overall, the total number of morphospecies represent approximately 25 percent of the total number of bees reported to occur in the High Plains of western Texas (approx. 280 species, unpubl. data).

Bee community data developed from this study was compiled by collecting bees using bee bowls (i.e., 3.5 oz plastic cups), which is a type of pan trap containing a soapy water solution. These traps are known to bias community samples because they variably detect species, with a known (high detection) bias towards the abundant small sweat bees (i.e. Halictidae). Therefore, while a high number of bee taxa collected in this study occurred with less than 5 individuals, this could reflect both taxon detection using bee bowls or their naturally low abundances. While the current data provides a strong relative snapshot of the community across farms, including other means of collecting (e.g. hunt sampling, belt transects, blue vane traps, etc.) should improve information on abundances and species’ rarity in these systems.

Table 3. Summary of bees collected from four farms in the Southern High Plains.

Overall, and as typically observed in native bee communities, there were a few taxa that numerically dominated the bee community (Table 4). The top 10 numerically dominant taxa (above upper dashed line) made up over 85 percent of the total number of individuals collected, with the top five taxa comprising over 72 percent of the total number of individuals. A total of forty-four morphospecies occurred with less than 10 individuals across all sampling dates and locations. Twenty-seven morphospecies occurred in abundances greater than 10 individuals (above lower dashed line), and only 14 morphospecies occurred with greater than 50 individuals.

Table 4 Bee morphospecies collected from this study, ranked by total numbers of individuals collected and taxon relative abundances.

Bee Community Relationships with Irrigation Scenarios and Soil Conditions

Across the 16 sampled transects among fields, the total number of specimens collected ranged from 26 to 534 individuals, and from five to 40 bee morphospecies (Table 5). The highest numbers of individual bees collected came from the CRP plot (DgC, Fig. 2), and this site also had 31 morphospecies, while the greatest number of morphospecies were collected from one of the tree-lined field corners (n = 40, Fig. 3). An average of 22 morphospecies were collected among sampled fields. A dryland, subsurface drip, and center field had relatively high numbers of morphospecies (30, 29, and 28) and with relatively high numbers of individuals collected. The alfalfa field was sampled on only two dates, and therefore the numbers for this field (i.e., 28 morphospecies) would likely be at the higher end if it had been sampled on every sampling date. The lowest number of morphospecies (10, 10 and 12) came from three dryland fields, while a sandy tree lined corner contained the lowest number of morphospecies at five; however, this site suffered from bee bowls being damaged and covered by blowing sand, rendering the samples damaged and incomplete.

While the number of morphospecies (S) showed some variation across fields (from 5 to 40), the community diversity indices E, H and D were not found to be statistically different across irrigation scenarios, nor were these diversity indices correlated with any of the soil condition variables (see below). Furthermore, indicator species analysis provided no significant indicator taxa significantly related to any of the irrigation transition scenarios.

Table 5. Summary of native bees occurring across fields according to irrigation scenarios (i.e. center pivot, subsurface irrigation and dryland). One Conservation Reserve Program field was sampled (DgC), and three additional field-edge plots containing woody trees were sampled (Ma 1T, Ma 2T, and H_site) that were proximal to production fields.

S = Richness = number of non-zero elements in row; “Total # morphospecies” in table

H = Diversity = - sum (Pi*ln (Pi)) = Shannon`s diversity index

D = Simpson`s diversity index for infinite population = 1 - sum (Pi*Pi)

       where Pi = importance probability in element i (element i relativized by row total)

Figure 2. CRP grassland site (DgC), which contained the second highest number of morphospecies (31) and the overall highest number of bee individuals (534) (Table 5).

Figure 3. Tree-lined corner of Ma 1T, which contained the highest number of bee morphospecies (40) and a high number of bee individuals (416) (Table 5).

Table 6. Soil variables measured in each field from 2019 to 2023. These variables were used in multivariate ordinations to analyze relationships between soil variables and bee communities.

Figure 4. NMDS ordination biplot showing each field (labelled) according to irrigation scenario (symbol). The proximity of fields in the graph reflect the general similarities in bee communities, with fields futher apart being more dissimilar in bee community structure. The “tree” sites and the alfalfa samples were included in this analyses, but these were not used in the analyses of soil variable relationships (i.e. soil was not sampled at these additional fields: H_site, Ma 1T, ans Ma 2T).  The final stree of this ordination was 8.55, and a 2-dimensional solution was recommended (i.e. most variation is extracted in just two orthogonal axes). The first two axes were significant based on Monte Carlo simulations using randomized data.

Figure 5. NMDS ordination with overlay of a second environmental matrix containing the soil variables in Table 6, using soil data from 2019 (top figure) and 2021 (bottom figure). The variables Soil C and F/B ratio were weakly correlated bee communities along axis 2 in 2021 (bottom), while in 2019 N-N was weakly correlated with sites grouped according to native bee communities along axis 1 (top).  

Figure 6. Principal components analysis (using correlation matrix) showing relationships of soil condition variables and S, E, H, and D community indices (Table 5) across fields different irrigation scenarios (symbols). Soil condition variables were highly multicollinear, with vectors correlated and grouped along axis 1. The top figure includes the CRP field that was heavily influencing the PCA biplot and site groupings (as an outlier; DgC). A second PCA (bottom) was run with this outlier removed, which resulted in a better separation of sites along axis 1, but still with high multicollinearity of variables along axis 1 and influenced by another field (DgA) located at the same farm as the CRP plot (DgC). DgA was also an initial grassland control containing a good stand of TEFF grass, but this field was terminated in 2022.

While no bee nests were observed in the field during our transect sampling or during opportunistic surveys, a large nest of the chimney sunflower bee (Diadasia enavata) was observed at the Quaker Avenue Research Farm at Texas Tech University. Since 2020, a total of three bee species have been observed in large nesting aggregations at this location (Fig. 7).

Figure 7. A soil nest of a native bee (shown by darker soil), Diadasia enavata. Students of the course Introductory Entomology at Texas Tech University had the opportunity to learn about the importance of undisturbed bare soil habitats on farms for native bee nesting, including old unused roadbeds as shown here.

(Click link) Video of Diadasia enavata nesting (Quaker Avenue Research Farm, Department of Plant and Soil Science, Texas Tech University)

Figure 8. Pinned specimens of the chimney sunflower bee, Diadasia enavata, collected from the large ground nest shown in Fig. 7 above.

Figure 9. A specimen drawer showing the organization of bee specimens prior to morphological analysis and data digitization.

Figure 10. A native bee cleptoparasite, Holcopasites stevensi. This bee was collected from Ma 1A and represents only a handful of specimens collected from Texas.

Figure 11. Small native bees in the genus Perdita. This group is very diverse with a total of 17 morphospecies identified from the SARE collections. Ongoing work continues to document morphological variation of this group to determine species level biodiversity. (Photos by undergraduate native bee researcher J. Winsauer)

Overall Findings of Pollinator Communities and Relationships with Irrigation Transitions and Soil Conditions

• The bee community occurring across the partner farms was diverse, represented by a total of 71 morphospecies including species rarely collected (e.g. Fig. 9).
• A few taxa dominate the bee community, with the top five numerically dominant taxa comprising over 72 percent of the total number of individuals collected. Most native bees were represented by less than 10 individuals.
• Bee communities were not different among irrigation transition scenarios, although there was a wide range of morphospecies numbers and total abundances across fields. Total abundances, morphospecies richness and diversity indices were not significantly different among transition scenarios or correlated with soil variables.
• The high number of morphospecies and total individuals in the CRP site, tree-lined corner (Ma 1T), and the house site (H_site) suggest these could be important on-farm habitats for providing foraging and nesting resources for native bees. A need exists to better understand how these unmanaged wild lands provide resources for bees and other beneficial insects, and how this biodiversity could assist sustainable agriculture.