Different currencies for calculating resource phenology result in opposite inferences about trophic mismatches

Shifts in phenology are among the key responses of organisms to climate change. When rates of phenological change differ between interacting species they may result in phenological asynchrony. Studies have found conflicting patterns concerning the direction and magnitude of changes in synchrony, which have been attributed to biological factors. A hitherto overlooked additional explanation is differences in the currency used to quantify resource phenology, such as abundance and biomass. Studying an insectivorous bird, Sanderling, and its prey, we show that the median date of cumulative arthropod biomass occurred, on average, 6.9 days after the median date of cumulative arthropod abundance. In some years this difference could be as large as 21 days. For 23 years, hatch dates of Sanderlings became less synchronized with the median date of arthropod abundance, but more synchronized with the median date of arthropod biomass. The currency-specific trends can be explained by our finding that mean biomass per arthropod specimen increased with date. Using a conceptual simulation, we show that estimated rates of phenological change for abundance and biomass can differ depending on temporal shifts in the size distribution of resources. We conclude that studies of trophic mismatch based on different currencies for resource phenology can be incompatible with each other.

Methods
To evaluate how our choice of currency for prey phenology will affect our perception of a phenological mismatch, we introduce a model system: the breeding phenology of an arctic shorebird, Sanderling (Calidris alba), and the phenology of their arthropod prey. We derive measures of arthropod phenology in terms of abundance and biomass (with the latter inferred using allometric length-biomass regressions). For this purpose, we analyzed 23 years of arthropod and bird data collected at Zackenberg. In addition, we use a simple simulation to assess how rates of phenological change are affected by the choice of currency for prey phenology. Since 1996 arthropod communities have been sampled throughout the snow-free season at Zackenberg, northeast Greenland (74°28' N, 20°34' W). To calculate annual seasonal trends in arthropod biomass at Zackenberg we first calculated the average number of specimens caught per taxonomic group per day per trap. This we did for each day of the time window during which a trap was active. These data only comprise counts of arthropods but do not contain measurements of the length or weight of individual specimens. Therefore, to calculate trends in biomass we first allocated a length to each specimen by sampling from an additional dataset containing taxon-specific length distributions. Once a length was allocated to each specimen, we calculated each specimen’s biomass using taxon-specific length-biomass regressions. For each year, we then computed daily estimates of total arthropod biomass by summing daily biomass estimates of all taxonomic groups and averaging them across all active traps. Using this adjusted dataset we then calculated the median date of cumulative arthropod numbers and the median date of cumulative arthropod biomass for each year. In addition, we assessed hatching dates of Sanderlings in Zackenberg by visiting nests or inferred this from egg flotation data or chick weights. We then calculated an annual estimate of mismatch by subtracting (I) the median date of arthropod abundance, and (II) the median date of arthropod biomass from the median hatch date of Sanderling chicks. We then assessed trends in the degree and direction of mismatch depending on the chosen currency by inspecting the slope of linear models. We used case-bootstrapping to calculate 95% quantile confidence intervals for all statistics. In addition, we used a simple conceptual simulation to assess how rates of phenological change might be impacted by the choice of currency when calculating resource phenology.