Organism abundance is a critical parameter in ecology, but its estimation is often challenging.Approaches utilizing eDNA to indirectly estimate abundance have recently generated substantial interest. However, preliminary correlations observed between eDNA concentration and abundance in nature are typically moderate in strength with significant unexplained variation.Here we apply a novel approach to integrate allometric scaling coefficients into models of eDNA concentration and organism abundance. We hypothesize that eDNA particle production scales non-linearly with mass, with scaling coefficients < 1. Wild populations often exhibit substantial variation in individual body size distributions; we therefore predict that the distribution of mass across individuals within a population will influence population-level eDNA production rates. To test our hypothesis, we collected standardized body size distribution and mark-recapture abundance data using whole-lake experiments involving nine populations of brook trout. We correlated eDNA concentration with three metrics of abundance: density (individuals/ha), biomass (kg/ha), and allometrically scaled mass (ASM) (∑(individual mass 0.73 )/ha). Density and biomass were both significantly positively correlated with eDNA concentration (adj. R 2 = 0.59 and 0.63, respectively), but ASM exhibited improved model fit (adj. R 2 = 0.78). We also demonstrate how estimates of ASM derived from eDNA samples in 'unknown' systems can be converted to biomass or density estimates with additional size structure data. Future experiments should empirically validate allometric scaling coefficients for eDNA production, particularly where substantial intraspecific size distribution variation exists. Incorporating allometric scaling may improve predictive models to the extent that eDNA concentration may become a reliable

17Organism abundance is a critical parameter in ecology, but its estimation is often challenging. 18Approaches utilizing eDNA to indirectly estimate abundance have recently generated substantial 19 interest. However, preliminary correlations observed between eDNA concentration and 20 abundance in nature are typically moderate in strength with significant unexplained variation. 21Here we apply a novel approach to integrate allometric scaling coefficients into models of eDNA 22 concentration and organism abundance. We hypothesize that eDNA particle production scales 23 non-linearly with mass, with scaling coefficients < 1. Wild populations often exhibit substantial 24 variation in individual body size distributions; we therefore predict that the distribution of mass 25 across individuals within a population will influence population-level eDNA production rates. To 26 test our hypothesis, we collected standardized body size distribution and mark-recapture 27 abundance data using whole-lake experiments involving nine populations of brook trout. We 28 correlated eDNA concentration with three metrics of abundance: density (individuals/ha), 29 biomass (kg/ha), and allometrically scaled mass (ASM) (∑(individual mass 0.73 )/ha). Density and 30 biomass were both significantly positively correlated with eDNA concentration (adj. R 2 = 0.59 31 and 0.63, respectively), but ASM exhibited improved model fit (adj. R 2 = 0.78). We also 32 demonstrate how estimates of ASM derived from eDNA samples in 'unknown' systems can be 33 converted to biomass or density estimates with additional size structure data. Future experiments 34 should empirically validate allometric scaling coefficients for eDNA production, particularly 35 where substantial intraspecific size distribution variation exists. Incorporating allometric scaling 36 may improve predictive models to the extent that eDNA concentration may become a reliable 37 indicator of abundance in nature. 38 39 3

1. Sustainable harvesting of wild populations relies on evidence-based knowledge to predict harvesting outcomes for species and the ecosystems they inhabit.Although harvesting may elicit compensatory density-dependence, it is generally size-selective, which induces additional pressures that are challenging to forecast. Furthermore, responses to harvest may be population-specific and whether generalizable patterns exist remains unclear. Taking advantage of Parks Canada's mandate to remove introduced brook troutSalvelinus fontinalis to restore alpine lakes in Canadian parks, we experimentally applied standardized size-selective harvesting rates (the largest ~64% annually) for three consecutive summers in five populations with different initial size structures. Four unharvested populations were used as controls.3. At reduced densities, harvested and control populations exhibited similar densitydependent increases in specific growth, juvenile survival and earlier maturation.However, size-selective harvesting simultaneously induced changes to size and age structure that contrasted among harvested populations. Average body length decreased in three of five harvested populations, whereas it tended to increase in control populations over the 3 years. We also detected contrasting, populationspecific changes in body length variability and ultimately in length-and age-atharvest in harvested populations but not controls.4. Overall, populations with smaller, more homogeneous body sizes, and living at high densities were most resilient to size-selective harvesting, exhibiting the smallest change in size-at-age. In contrast, large-bodied populations exhibited more substantial size-structure changes following selective harvesting: largebodied populations experienced either stabilizing or disruptive pressures, when initial length variability was high or low, respectively. Synthesis and application.Our results show that within species, size-selective harvesting inherently leads to more risk and uncertainty when harvesting populations with larger and more varied body sizes than smaller-bodied populations with less range in body size. Our study supports prioritizing regulations that protect | 1303