Food availability varies considerably over space and time in wetland systems, and consumers must be able to track those changes during nutrient demanding points in the life cycle like breeding. Resource tracking has been studied frequently among herbivores, but receives less attention among consumers of macroinvertebrates. We evaluated the change in resource availability across habitat types and time, and the simultaneous density of waterfowl consumers throughout their breeding season in a high-elevation, flood-irrigated system. We also assessed whether the macroinvertebrate resource density better predicted waterfowl density across habitats, compared to consistency (i.e., temporal evenness) of the invertebrate resource or taxonomic richness. Resource density varied marginally across wetland types but was highest in basin wetlands (i.e., ponds) and peaked early in the breeding season, whereas it remained relatively low and stable in other wetland habitats. Breeding duck density was positively related to resource density, more so than temporal resource stability, for all species. Resource density was negatively related to duckling density, however. These results have the potential to not only elucidate mechanisms of habitat selection among breeding ducks in flood-irrigated landscapes, but also suggest there is not a consequential trade-off to selecting wetland sites based on energy density versus temporal resource stability and that good-quality wetland sites provide both.

Macroinvertebrate Data Collection

We collected nektonic invertebrate samples using 2-L activity traps in 2020 and 2021 (Murkin et al. 1983). We placed traps at randomly-selected points within 40 wetland sites that encompassed the five different wetland habitats. The sites we sampled in four of the wetland habitats (basin wetlands, riparian wetlands, irrigation ditches, and flooded hay meadows) spanned three individual properties, including two private ranches and Arapaho NWR (Figure 1). We also sampled three public reservoirs across the study area. We selected three random wetlands of each type on each of the three properties, and three random sampling points within each selected wetland with the exception of reservoirs. We randomly selected two 200-m length plots of shoreline in each reservoir and three random points within each plot to sample (Cooper and Anderson 1996, de Szalay et al. 2003, Behney 2020). In total, we deployed 126 traps during each sampling occasion (3 properties x 4 wetland habitats x 3 sampled wetland sites of each variety x 3 sampled points in each wetland) = 108 samples, + (3 reservoirs x 2 plots in each reservoir x 3 points in each plot = 18 samples). Traps remained in the wetlands for 48 hours every fourteen days, resulting in six sampling occasions each year over the course of the breeding season (13 May through 22 July). No traps were placed if the wetland was dry on a given sampling occasion, and that status was noted and aquatic invertebrate resource density was treated as a zero for that occasion. In the event that a wetland was not flooded or less flooded during the second year of sampling (2021), we randomly selected new points within the new boundary of the same wetland and treated those sampling points as unique from the original locations, but nested within the same site. Occasionally traps became dislodged and either went missing or floated to the surface, in which case we replaced traps and allowed them to remain in the wetland for the subsequent 48 hours.

Activity traps had a 15 cm opening at the widest part of the funnel and a 2 cm opening at the narrowest part of the funnel. We placed them so that the top of the widest part of the funnel was approximately 1 cm above the surface of the water to capture invertebrates in the part of the water column in which dabbling ducks most often forage (Guillemain et al. 2000, Behney 2020). Upon collection, traps were emptied into a mesh sieve-bottom bucket. All individual invertebrates from the sample were placed into plastic storage cups and stored in 70% ethanol until processing. We emptied each sample into a 0.355 mm (number 45) gauge mesh sieve in a wet lab and moved all individuals to a Petri dish for identification and counting (Behney 2020). We placed samples under a dissecting microscope (AmScope SM-1BSY-64S Stereo Zoom Microscope) and identified individuals to taxonomic Family when possible. Any sample containing more than 1000 individuals of a given Family was subsampled by 16.6% using a 6 x 6 square gridded Petri dish (Williams et al. 2014, Behney 2020). We counted individuals in a random subset of six of the 36 cells and multiplied by six to estimate the total number of individuals of that Family in the sample.

Waterfowl Data Collection

We conducted breeding pair counts of ducks on the same wetland sites being sampled for macroinvertebrates (n=40) using a dependent double-observer methodology during the breeding seasons of 2020 and 2021 (Nichols et al. 2000). Pair count survey timing coincided with the first three macroinvertebrate sampling occasions. A primary observer counted every individual dabbling and diving duck observed and reported the number to a secondary observer, who recorded data while also recording any observations missed by the primary observer (Roy et al. 2021). We restricted the dataset to the four most common species of breeding ducks in our study system, which included cinnamon teal (Spatula cyanoptera), gadwall (Mareca strepera), mallard (Anas platyrhynchos), and lesser scaup (Aythya affinis). Although we used the standard pair count practice of separating lone drakes from paired ducks in each count to identify breeding phenology and thus the timing of our first invertebrate sampling occasion, we used the total count of breeding ducks for the purposes of evaluating the relationship between duck density and invertebrate availability. In addition, few individuals were missed by the primary observer (i.e., detection probability was high), so we pooled observed drakes and hens of a given species to give us a site- and occasion-specific count.

Brood surveys also occurred on the sites sampled for macroinvertebrates, but followed an independent double-observer methodology during 2020-2021 (Nichols et al. 2000, Vrtiska and Powell 2011). Both observers counted the number of ducklings they observed and subsequently compared observations to determine whether they had been observing the same brood and compare the number counted. The smaller number of ducklings commonly observed at one time allowed for accurate count comparisons between observers (Pagano and Arnold 2009). The timing of counts coincided with the latter three macroinvertebrate sampling occasions. Observers counted all ducklings within a given brood and identified their age class according to Gollop and Marshall (1954). Observers spent a minimum of ten minutes at each wetland site and conducted surveys using window- or tripod-mounted spotting scopes and binoculars to allow time for hidden broods to become visible (Pagano and Arnold 2009, Walker et al. 2013). We restricted the dataset to ducklings of the same four duck species listed above, and pooled the total number of ducklings for a given species, site, and occasion.