Forest defoliator outbreaks alter nutrient cycling in northern waters

The study investigates how outbreaks of insect defoliators alter carbon (C) and nitrogen (N) cycling from forested catchments to lakes. Freshwaters reflect their catchments, with dissolved organic carbon (DOC) affecting lake light, thermal regime, toxicity, and food webs, and dissolved inorganic nitrogen (DIN) limiting primary production in many northern lakes. Prior work has emphasized wood-boring insects (e.g., bark beetles) that kill trees and increase organic matter inputs, but the effects of defoliators—which remove foliage without killing trees—on aquatic C and N fluxes are poorly understood. Defoliation can reduce foliar litter inputs while producing N-rich, biolabile frass, potentially decreasing DOC and increasing DIN in receiving waters and shifting N:P ratios. The authors hypothesize that defoliator outbreaks reduce lake DOC and increase DIN, with effects modulated by forest composition, and that these within-year changes can rival or exceed broader inter-annual trends. They test these ideas using 32 years of water chemistry and insect outbreak data across 12 boreal catchments in Ontario, Canada.

Previous studies on forest disturbance impacts on aquatic systems have focused on wood-boring insects that cause tree mortality, leading to increased litter inputs and reduced N uptake, often elevating DOC and DIN downstream. In contrast, defoliators consume foliage during early to mid-summer, potentially reducing autumn litter-derived C and N while depositing frass that is N-rich and biolabile due to inefficient N assimilation by larvae and pre-senescence foliar chemistry. Frass from deciduous leaves can have lower C:N than corresponding litter, enhancing microbial processing. Phosphorus in lakes is less tightly linked to forest litter inputs and more influenced by atmospheric deposition, wetland pathways, and internal cycling, so defoliation may shift N:P ratios without changing P. Tree species differences in foliar chemistry imply that catchment composition (deciduous vs coniferous) may modulate the biogeochemical fate of frass and litter.

• Study design: Analysis of 32 years (1985–2016) of monthly ice-free (May–October) lake water chemistry and annual insect outbreak data across 12 single-lake catchments (18–1045 ha) in Ontario, Canada (Algoma, Greater Sudbury, Muskoka, Temiskaming). All lakes are nutrient-poor; most are thermally stratified.
• Insect outbreak data: Annual aerial surveys by Canadian Forest Service and Ontario MNRF mapped moderate to severe defoliation across 439,661 km^2 of Ontario. For each study catchment, polygons of annual insect damage (defoliators including gypsy moth, forest tent caterpillar, spruce budworm, jack pine budworm, aspen two-leaf tier, Bruce spanworm) were clipped to boundaries, and percent damaged area computed.
• Forest cover: Monthly leaf area index (LAI) per catchment (1985–2016) derived by converting smoothed Landsat NDVI (30 m, 16-day; TIMESAT Savitzky–Golay smoothing) to MODIS LAI via a non-linear calibration using 2016 data (cloud-filtered MODIS/006/MCD15A3H; NDVI downscaled to match 500 m). Catchment proportions of purely deciduous and coniferous forest stands were extracted from Ontario Land Cover Compilation v2.0.
• Lake chemistry: Monthly surface grab samples near maximum depth measured DOC (after acidification, inorganic C removal, oxidation to CO2, colorimetry), DIN (nitrate + ammonium by automated colorimetry), and TP. When multiple samples per month existed, values were averaged.
• Statistical analysis: For LAI, DOC, DIN, TP, and DIN:TP, separate linear mixed-effects models (nlme) were fitted with fixed effects for percent catchment defoliated, month (May–Oct), and their interaction; forest composition interactions (percent damaged × deciduous and × coniferous) and main effects of forest cover were tested. Catchment identity was a random effect; temporal autocorrelation was modeled with a continuous first-order autoregressive structure (CAR(1)) within lakes. Estimated marginal means and 95% CIs were computed with emmeans.
• Contextual comparison with broader trends: Within-year effects of complete catchment defoliation on annual ice-free mean DOC (and DIN) were compared to between-year trends. For DOC, absolute Theil–Sen slopes (1990–2016) from 266 boreal and north temperate lakes (ICP Waters; relatively undisturbed) were contrasted with modeled outbreak effects. For DIN, between-year trends were computed for the 12 study lakes (ICP Waters lacked DIN).
• Regional outbreak frequency: Using Ontario Integrated Hydrology catchments for all lakes ≥5 ha, the number of years with ≥50% catchment defoliated (all defoliator species) was counted between 1990 and 2016 to assess frequency and pervasiveness (overlapping polygons merged to avoid double counting).

• Canopy impacts:
  • LAI decreased during outbreaks, consistent with defoliation: mean 22% lower LAI across catchments as percent damaged increased from 0 to 100% during the growing season; peak decrease 24% in July. No significant LAI differences in May, September, or October between outbreak and non-outbreak conditions.
• DOC responses:
  • During years with highest insect damage, DOC decreased by a mean 0.71 mg L−1 (95% CI: 0.19–1.23) from July onward; relative reduction 19–24% compared to no disturbance. With ~45% catchment disturbed, DOC reduction was 0.32 mg L−1 (95% CI: 0.09–0.56).
  • Seasonal pattern: In non-outbreak years, DOC increased from May to July and stayed elevated to October; this seasonal rise was absent when ≥47% of the catchment was disturbed.
  • Forest composition effect: DOC declines with defoliation were stronger in catchments with more deciduous stands. At median deciduous cover (0.26), 0→100% damage reduced DOC by 0.48 mg L−1 (95% CI: 0.08–0.88); at upper quartile (0.60), reduction was 1.13 mg L−1 (95% CI: 0.28–2.00), a 32% relative decline. Coniferous cover did not modify DOC effects.
• DIN responses:
  • At 100% catchment disturbance, DIN increased by 0.03 mg L−1 (95% CI: 0.01–0.04) in July, persisting through the ice-free season with mean increase 0.03 mg L−1 (95% CI: 0.01–0.04). Relative increase peaked at 134% in October when damage rose from 0 to 100%.
  • DIN always showed a sharp May→July decrease and a September→October increase, irrespective of defoliation.
  • No modifying effect of forest composition on DIN.
• Phosphorus and stoichiometry:
  • Total P showed no change with outbreaks; thus N:P ratios increased during outbreaks.
• Magnitude relative to broader trends:
  • Outbreak-driven within-year DOC decrease (annual ice-free mean) was 0.16 mg L−1 (95% CI: 0.01–0.32) over 1990–2016, exceeding 85% of absolute inter-annual DOC trends in 266 lakes and double the mean absolute DOC trend (0.08 mg L−1 yr−1).
  • Outbreak-driven annual DIN increase was 0.012 mg L−1 (95% CI: 0.001–0.021), exceeding all annual DIN trends observed in the 12 study lakes and 12× the mean absolute annual DIN change.
• Outbreak frequency and extent:
  • Across 60,430 Ontario lake catchments (≥5 ha), median of 4 outbreak years per catchment during 1990–2016; 80% of catchments had at least one outbreak. Half of surveyed catchments experienced defoliator outbreaks at least every 5 years.

The findings support the hypothesis that defoliator outbreaks reduce lake DOC and increase DIN, with effects shaped by forest composition and peaking during mid to late growing season. Mechanistically, defoliation reduces canopy leaf area and autumn litter inputs while converting foliage into N-rich, highly biolabile frass. This frass primes microbial activity in soils and aquatic systems, increasing respiration and microbial uptake of labile C and thereby lowering DOC exported to lakes. Inefficient N assimilation by larvae and production of soluble N-rich frass elevate DIN in receiving waters, with no parallel increase in P, raising N:P ratios. Stronger DOC reductions in deciduous-dominated catchments align with higher biolability of deciduous foliage; defoliators did not preferentially damage forest types, but foliar quality mediated biogeochemical outcomes. Compared to wood-boring insect outbreaks that increase DOC via tree mortality and litter accumulation, defoliator outbreaks produce contrasting DOC dynamics while still elevating DIN through different pathways. The within-year outbreak effects were large relative to long-term regional trends, indicating that frequent, widespread defoliation can temporally mask or modulate background trajectories in lake biogeochemistry. These nutrient shifts have implications for lake food webs: reduced DOC may increase light availability, while elevated DIN and unchanged P can shift N:P ratios, potentially favoring heterotrophic bacteria over algae under P limitation and altering productivity timing, especially in late autumn during stratification. Climate change may amplify these dynamics by increasing outbreak frequency and expanding deciduous cover, enhancing the biolability of inputs and the magnitude of aquatic responses.

This study demonstrates that insect defoliator outbreaks are a pervasive, recurrent driver of lake biogeochemistry in northern forests, causing substantial within-year decreases in DOC and increases in DIN that can exceed or mask broader inter-annual trends. Effects are strongest in deciduous-dominated catchments and occur without concomitant changes in total P, shifting N:P ratios. The results underscore the need to incorporate defoliator disturbances into regional and global C and N budgets and aquatic ecosystem models. Future research should quantify how algal versus bacterial production responds during and after outbreaks, track changes in community composition with altered N:P, and assess how climate-driven changes in forest composition and insect dynamics will modulate land–water nutrient transfers.

• Observational design using aerially mapped defoliation captures moderate to severe events; smaller-scale or low-severity defoliation may be underrepresented.
• LAI was inferred by converting smoothed Landsat NDVI to MODIS LAI using a non-linear calibration from a single year (2016), introducing potential uncertainty.
• Lake chemistry was sampled monthly during ice-free periods only via surface grabs; sub-monthly dynamics and under-ice conditions were not assessed.
• The broader 266-lake dataset lacked DIN measurements, limiting inter-annual DIN comparisons to the 12 study lakes.
• Frass production and export were not directly measured; mechanisms are inferred from established literature and observed water chemistry patterns.
• Mixed and sparse forest classes were excluded from forest composition analyses, potentially simplifying stand heterogeneity effects.