Over two years across southern Alberta 17 alfalfa seed fields (Figure 1) were sampled for insect pests (20-sweep samples, 38.1 cm diameter net, 180° sweeps) every 1-2 weeks between late May and late August in 2023 and 2024. Four locations within each field at each field corner had two plots each with insecticide treated and untreated plots (Figure 2). Sweep samples were taken from the centre of plots to avoid edge effects when temperatures were above 15°C during the day, samples were collected in plastic bags that were stored at -20°C. Samples were sorted and key insect pests and their life stages were counted, including alfalfa weevil (Hypera postica) larvae (1st/2nd and 3rd/4th instars) and adults, Lygus spp. adults and instars (1-5th instar), and alfalfa plant bug (Adelphocoris lineatus). For different pests, the life stage is important given the differences in feeding behaviour, for example alfalfa weevil 3rd and 4th instars are more damaging that younger instars and adults.

Weather station data obtained from either Environment and Climate Change Canada or Alberta Climate Information Services (ACIS) for daily temperatures (mean, max, min), daily precipitation, and the summed daily precipitation from the past 7 days.  Daily maximum and minimum temperatures were used to calculate degree days for alfalfa weevil (base 9°C, maximum cap at 31°C), see equation below. 

GDD = ((TMAX+TMIN)/2 )−Tbase

In addition, we collected information on plot yield, plant stand density, stem density, and plant stand age. Plant characteristics are part of management practices as a proxy for seeding density, irrigation, and how long growers keep a field based on their farming contracts. Farmers treated their fields with insecticides as they deemed necessary, insecticide treated plots in this study reflect farmer's on-farm decision making to maximize yield while considering pollinators. Data was collected on insecticide applications, including active ingredients/chemistry groups, number of applications, and spray dates for all plots within a field. 

Field characteristics and insecticides impacts on insect pest communities

For each plot within each field, the average insect life stage counts were calculated and separated into two communities matrices, one for untreated plots and one for insecticide treated plots. These community matrices were separately used in Multivariate Correlation and Regression Tree (MRT) analyses to examine how field and management practices influence insect pest communities. Plant stand characteristic variables included plant stage age, plant stand density, and stem density, while insecticide management variables included number of alfalfa weevil insecticide applications, and number of other insecticide applications. For the untreated MRT only plant stand age, plant stand density, and stem density were use as explanatory variables. Year was included as a parameter in both MRT analyses, to account for the low pressure year in 2023 and to avoid spurious interpretations due to the large number of 2nd year fields in 2023. Insect community data was centred and scaled prior to MRT analyses to better visualize and interpret dendrograms.

Seasonality of insect pests and management practices

Growing Degree Days (GDD) for alfalfa weevil (base 9°C) were calculated. Temporal plots were used to visualize insect pest populations in relation to growing degree days, recommended economic thresholds and insecticide applications. Average differences between application dates and the upper GDD recommendation of 260 GDD were calculated. Miridae insects were grouped together as producers determine economic thresholds from combined numbers and seasonal timings were compared for each year, using t-tests between early (May to mid-July) and late (mid-July to August). 

Clustering of insect pest pressures

Fields were grouped into different clusters based on pest pressures (early and late season averages for alfalfa weevil larvae, alfalfa plant bug, Lygus bug adults and nymphs) from unsprayed plots using the hclust package in R. Ward Agglomerative Hierarchical Clustering method groups smaller clusters into larger clusters, the insect pest community matrix in this study was transformed using a Hellinger transformation and a Euclidean distance was used for the clustering. Four insect pest clusters were identified by inspecting silhouette distances. Pest classes were visualized using a non-metric multidimensional scaling ordination, computed using the ecodist package in R with 80% confidence ellipses plotted using the ordiellipse function from the vegan package in R. To test for differences in classes a permANOVA was conducted (p-value=0.06) using the adonis2 function from the vegan package in R. To qualitatively describe classes, follow-up ANOVAs were conducted for each insect pest and pairwise tests with a Dunn-Šidák multiple comparison correction. For percent yield differences and for unsprayed yields, ANOVAs were conducted to compare pest classes, followed by pairwise tests with a Dunn-Šidák multiple comparison correction.

Insect pests and alfalfa seed yield

To estimate seasonal pest pressure without biasing results due to differing insect phenology, early (between May and 15 July) and late (after July 15) season averages for insect pests were used (alfalfa weevil adults, alfalfa weevil small larvae, alfalfa weevil large larvae, Lygus adults, total Lygus nymphs, and alfalfa plant bugs). These averages were centred and scaled and used as fixed effects of alfalfa seed yield. Year and field were included in the model as random effects. Multicollinearity was checked using a variance inflation factor (VIF), which removed alfalfa weevil adults early season alfalfa plant bugs and late season alfalfa weevil parameters to keep VIFs below 5. Models were computed for all insect pests and selected based on the lowest Akaike Information Criterion (AIC).  Model assumptions and residuals were examined using the performance package in R. 

All analyses conducted in this study were performed in R version 4.4.1 , except for regression tree analyses that used R version 4.2.3. [13].