1 INTRODUCTION

Organismal responses to climatic warming often include phenological shifts in major life-history events, which can lead to heterogeneous responses among different functional groups within food webs (Thackeray et al., 2016). Small-bodied ectothermic organisms at lower trophic levels often respond with stronger phenological adjustments than large endothermic organisms at higher trophic levels, which can lead to a trophic mismatch between consumers and their food sources (Both et al., 2009; Cohen et al., 2018; Kharouba & Wolkovich, 2023). Arctic ecosystems are currently warming two to four times faster than the rest of the earth (IPCC, 2021; Rantanen et al., 2022). Hence, arctic food webs are especially likely to be subject to trophic mismatches between consumers and their resources (Post et al., 2009; Schmidt et al., 2017).

Insectivorous migratory birds breeding in the Arctic are expected to be especially prone to warming-induced mismatch (McKinnon et al., 2012; Miller-Rushing et al., 2010; Saalfeld et al., 2019; but see Corkery et al., 2019; McKinnon et al., 2013). Arthropods are resident ectotherms, and their phenology is strongly affected by local environmental conditions (Culler et al., 2015; Høye & Forchhammer, 2008a; Shaftel et al., 2021). By contrast, the migration and breeding phenology of some arctic birds can be affected by a wide range of cues and environmental conditions encountered en route to and at breeding sites (Liebezeit et al., 2014; Smith et al., 2010; Ward et al., 2016; Winkler et al., 2014). Arctic birds also have a relatively short window of time for reproduction, and food availability during the chick rearing period is especially critical as it may affect fitness through juvenile growth and survival (McKinnon et al., 2012; Reneerkens et al., 2016; Saalfeld et al., 2019, 2021). Chicks of insectivorous birds mostly feed on active and visible arthropods (Richards & Gaston, 2018; Schekkerman et al., 1998) and their availability is typically characterized by a relatively short summer pulse (Bolduc et al., 2013; Danks, 2004; Tulp & Schekkerman, 2008). Consequently, the response of arctic arthropods to warming can strongly affect the likelihood of trophic mismatch.

Studies based on time series are often used to predict long-term effects of warming on wildlife species and trophic interactions (Kharouba et al., 2018; Parmesan, 2007). However, empirical studies of the phenological responses of arctic arthropods to a prolonged increase in summer temperatures are rare and typically limited to single-site assessments (e.g., >25 years time series; Høye et al., 2021). Such studies showed that arctic arthropod assemblages can change relatively quickly when exposed to warmer temperatures for several years (e.g., changes in the relative abundance of functional groups; Koltz et al., 2018), which can cause cascading effects on arthropod availability for consumers (Schmidt et al., 2017).

Assuming that the variation of a parameter through space can also be used to predict its variation through time, a space-for-time substitution approach has the potential to improve our assessment of the long-term effects of warming on ecological communities (Blois et al., 2013; Elmendorf et al., 2015; Pickett, 1989). Although the approach has some important caveats, it can nonetheless be useful when the drivers that control biological processes are the same drivers that operate in space and if community dynamics and assemblages can respond relatively quickly to persistent environmental changes (Damgaard, 2019; Wogan & Wang, 2018).

Here, we apply a space-for-time substitution approach to improve our ability to predict the general response of arctic arthropods to warming, and hence the potential consequences on consumers like migratory birds. Our study is based on a pan-arctic dataset of surface-active and low-flying arthropod biomass collected from the Subarctic to the extreme High Arctic. It spans a large temperature gradient (>10°C in June and July average daily air temperature) overlapping the expected temperature increase in the Arctic and allows comparisons between species assemblages shaped by long-term exposure to warmer temperatures. Temperature is known to affect arthropod emergence phenology (e.g., date of peak biomass; Figure 1a) but also the magnitude of peak arthropod biomass (Figure 1b; e.g., Bolduc et al., 2013; Tulp & Schekkerman, 2008). The consequences of warming on the availability of arthropods for Arctic-nesting birds are likely to depend on the strength of such combined effects (Figure 1c). Despite potential differences in temperature responses among arthropod species, we hypothesized that higher summer temperatures would be associated with earlier peak arthropod biomass and higher arthropod biomass available to Arctic-nesting insectivorous birds across the temperature gradient covered by our study sites.

2 METHODS

2.1 Arthropod data

2.1.1 Field sampling

Arthropods were sampled during the breeding season of insectivorous birds (approximately June to August) at 19 field sites distributed across most Arctic and Subarctic bioclimatic zones (Figure 2; Leemans, 1992; Walker et al., 2005). Each site was sampled for 1–19 years, and over a period of 26–143 days per year (medians = 3 years and 56 days, respectively; Table 1), covering the estimated period of peak arthropod biomass. At each site, 6–20 traps were deployed and typically emptied every 1–3 days, except for Zackenberg, Hochstetter Forland, and Chipp River, where sampling was performed once a week. The biomass estimates were then converted to daily values as described below.

TABLE 1. Arctic and Subarctic study sites ranked by the average June and July daily air temperature of years included in our analyses (ERA Interim; Dee et al., 2011).

Site name	Mean summer, T (°C)	Coordinates	Sampling years (n)
Utqiaġvik	1.2	71°18′ N, 156°45′ W	2010–2016 (7)
Alert	1.5	82°30′ N, 62°21′ W	2007–2008 (2)
Zackenberg	1.6	74°28′ N, 20°34′ W	1998–2016 (19)
Hochstetter Forland	2.0	75°09′ N, 19°42′ W	2011–2014, 2016–2017 (6)
Ikpikpuk	2.2	70°33′ N, 154°43′ W	2010–2012 (3)
Medusa Bay	2.4	73°20′ N, 80°32′′  E	1996, 2000–2002 (4)
Prudhoe Bay	2.6	70°12′ N, 148°27′ W	2010 (1)
Bylot Island	3.1	73°80′ N, 79°58′ W	2005–2017 (13)
Chipp River	3.2	70°41′ N, 155°18′ W	2013 (1)
Canning River	3.5	70°26′ N, 145°51′ W	2010–2012 (3)
Igloolik	3.7	69°24′ N, 81°48′ W	2014, 2017 (2)
Colville	3.8	70°26′ N, 150°41′ W	2011–2012, 2014–2017 (6)
Herschel Island	4.1	69°35′ N, 138°55′ W	2007–2008 (2)
Mackenzie Delta	4.9	69°22′ N, 134°53′ W	2011–2012 (2)
Southampton Island	5.0	63°59′ N, 81°40′ W	2006–2008, 2010–2012 (6)
Nome	8.3	64°27′ N, 164°58′ W	2010–2012 (3)
Churchill	8.4	58°45′ N, 94°04′ W	2010–2011 (2)
Cape Krusenstern	9.7	67°06′ N, 163°29 W	2011–2012 (2)
Akimiski Island	11.7	53°00′ N, 81°20′ W	2009 (1)

• Note: For a map including bioclimatic zones, see Figure 2. For detailed weather information, see Appendix S4.

At each study site, arthropods were sampled in dry uplands and low wetlands, which were the two main habitats used by the common species of shorebirds, passerines, and other insectivorous birds during their chick rearing period. Arthropods were collected in open areas within each habitat using modified Malaise traps with rectangular white pitfall traps (38 cm × 5 cm) at most sites (Bolduc et al., 2013; Brown et al., 2014), except Medusa Bay, Hochstetter Forland, and Zackenberg, where round white or yellow pitfall traps were used (10–11 cm in diameter; Schmidt et al., 2016; Tulp & Schekkerman, 2008). Therefore, we compared results obtained from these two trapping techniques conducted simultaneously at the same site before combining datasets (see below). Both techniques are passive traps and thus measure a combination of the abundance and the activity of surface-active and low-flying arthropods (Southwood & Henderson, 2000). Variation in biomass was used as a proxy of arthropod availability for surface-feeding insectivorous birds such as shorebirds (Bolduc et al., 2013; Kwon et al., 2019; McKinnon et al., 2012).

Arctic-breeding birds are likely to be gape-limited and restricted to prey that they can swallow whole. We excluded bumblebees and butterflies, as they were likely too large to be consumed by chicks (Kwon et al., 2019; Saalfeld et al., 2019; Schekkerman & Boele, 2009). We also excluded springtails (Collembola) and mites because they were a negligible part of the sampled biomass and were considered too small to be important prey for chicks (Bolduc et al., 2013; Ridley, 1980; Tulp & Schekkerman, 2008). The biomass of all remaining arthropods was pooled, as arctic insectivorous birds typically consume a broad diversity of species during the breeding season (Flemming et al., 2022; Wirta et al., 2015). The dry biomass for each trap was measured directly or estimated with equations using the length of the specimen to convert abundance of individuals to dry mass (see Appendix S1 for details). Wet and dry habitats were pooled to obtain a general index of arthropod availability (McKinnon et al., 2012). A standardized daily arthropod availability index (mg/trap) was calculated for each year and site, achieved by dividing the overall arthropod biomass by the number of sorted traps and the days elapsed between sampling events (Bolduc et al., 2013). Our index of arthropod abundance and activity accounted for the variable number of traps and the differences in sampling frequency across sites.

2.1.2 Data standardization

We carried out a calibration experiment because two different trapping techniques were used across our study sites (i.e., different colors, shapes, and sizes of the pitfall traps). We used both types of traps simultaneously at Bylot Island in 2018 to develop a linear regression linking the variation in the biomass of arthropods captured with the two different types of traps (see Appendix S2).

2.1.3 Arthropod phenology and biomass

Based on seasonal changes in the index of arthropod daily availability, we calculated three parameters for each year and field site: peak biomass, date of peak biomass (hereafter peak date), and seasonal biomass. Peak biomass was defined as the single highest recorded value of daily arthropod availability (mg/trap/day) and peak date was the rounded median date of the sampling period during which the peak biomass was observed. We explored alternative methods (generalized additive models and moving averages) and found similar results. Zackenberg, Hochstetter Forland, and Chipp River data were excluded from peak date analyses, because we considered that the weekly sampling frequency would strongly reduce the precision of the estimates compared to other sites that were sampled every 1–3 days. The daily index of arthropod availability can vary substantially before and after the peak date, and the period characterized by relatively high biomass can be relatively narrow or wide, depending on study years and sites (Bolduc et al., 2013; Saalfeld et al., 2019). To better describe the arthropod biomass available to birds, we also calculated the seasonal biomass, which is the cumulative biomass sampled during a 3-week period centered on the peak date (i.e., sum of daily biomass recorded during the period corresponding to the peak date ±10 days). Shorebird chicks are more vulnerable to starvation or marked reduction in growth rate during their first 9 days (McKinnon et al., 2012; Pearce-Higgins & Yalden, 2003). Our seasonal biomass estimates provided a proxy of food availability just before and after peak date, over a biologically significant period. Estimating seasonal biomass over a longer period generated similar results and the correlation between seasonal biomass estimates using a 21-day or 31-day window was high (r = .99, p < .001, n = 57). As the total sampling season was shorter at some study sites, the use of a 21-day window increased our sample size. Nevertheless, in nine out of 84 site-years, we were unable to estimate a seasonal biomass value using a 21-day window, as the period of sampling did not fully cover the minimum time window around the peak date. For 4 of the 9 cases where only 3 days were missing, the window was shifted up to 3 days earlier or later to include a full 21-day window of cumulative biomass (site and year: Canning in 2011, Mackenzie in 2011, Southampton in 2007, and Utqiaġvik in 2010).

3 RESULTS

A total of 16,134 arthropod traps were sampled at 19 different study sites for 1–19 years at each site, for a total of 84 site-years. Weather conditions varied substantially between sites (Figure 3; Appendix S4). The cumulative thawing degree-days between June 5 and July 15 varied from 15 DD to 426 DD.

The peak biomass value per site, averaged over years, ranged from 6 to 433 mg/trap/day, while the average seasonal biomass (assessed over the 21-day period centered on the peak date) varied from 77 to 4354 mg per site. Peak biomass and seasonal biomass were highly correlated (r = .96, p = .001, n = 75 site-years). Arthropod peak dates varied substantially within and among sites (50 days range; Figure 3a), but site-averaged peak dates occurred within the 1-month period between June 17 and July 18 across sites. In some cases, the intra-site range in peak dates was almost as large as inter-site variation (Figure 3a). For instance, peak dates were spread over a 24-day period at Bylot Island, a High Arctic site with a 13-year time series. This period included the average peak dates observed at 10 out of the 16 study sites where peak dates were estimated. Hence, in some years, birds nesting at Bylot Island experienced a range of arthropod phenology that was normally observed in much warmer or colder breeding sites.

Variation in arthropod peak dates was best explained by linear effects of DD and snowmelt (SN; Table 2). The top model indicated that an increase of 25 DD advanced the peak date by 1 day on average (b = −0.04, 95% CI = −0.07 to −0.02; Figure 3a; Table 2a). Peak date was 1 day earlier, when snow-free date was 4 days earlier, on average (b = 0.25, 95% CI = 0.06–0.43). Unlike the linear effect of DD found for peak date, the variation in peak biomass and seasonal biomass were best explained by a segmented effect of DD (Table 2). We found that DD had a positive effect on peak biomass but only below a threshold of 106 DD (95% CI = 64–148 DD; Figure 3b). Below this threshold, an increase in 25 DD generated on average a 43.5 mg/trap/day increase of peak biomass (b = 1.74, 95% CI = 0.24–3.24). Above 106 DD, the effect was not significant (b = −0.07, 95% CI = −0.35–0.22). Similarly, the positive effect of DD on seasonal biomass was detected only below a threshold of 177 DD (95% CI 89–266 DD; Figure 3c). Below this threshold, a rise of 25 DD increased seasonal biomass by 270 mg on average (b = 10.8, 95% CI = 4.22–17.38) while the relationship was not significant above the threshold (b = −2.28, 95% CI = −8.20–3.64). Excluding extreme values observed at one site (Herschel) had a marked influence on the parameter estimate but did not affect the main patterns (peak biomass: b = 0.75, 95% CI = 0.25–1.26, threshold = 133 DD, 95% CI = 73–194 DD; 2nd segment: b = 0.04, 95% CI = −0.14–0.22; seasonal biomass: b = 7.77, 95% CI = 2.28–13.25, threshold = 134, 95%CI = 59–209; 2nd segment: b = 1.46, 95% CI = −0.49–3.41).

3.1 Model illustration

Based on our empirical results, we created two alternative scenarios to visualize the potential effects of a temperature increase on arthropod availability for consumers, one above and one below the temperature threshold identified by our segmented biomass models (Figure 4). We created scenarios for (i) a relatively cold, generic Arctic site (current average of 50 DD; Figure 4a) versus (ii) a relatively warm, generic Low Arctic site (current average of 320 DD; Figure 4b). In both cases, we illustrated a temperature increase of 80 DD (equivalent to a 2°C increase in average daily air temperature between June 5 and July 15). A change of 80 DD allowed us to stay within the temperature range covered by each segment of the regression (below and above the breakpoint). For the generic cold site, we superimposed hatching periods of the three most common species at three of our cold study sites as points of reference (Bylot, Ikpikpuk, and Utqiaġvik; Lanctot et al., 2016). We repeated the same steps for the generic warm site, using data from three of our warm sites (Churchill, Cape Krusenstern, and Nome; Lanctot et al., 2016). In both scenarios, the model predicted a 3.2-day shift in average arthropod peak date (Figure 4a,b). Peak dates would occur during the current shorebird hatching periods recorded at representative study sites, except for the latest breeding species (especially for American Golden-plovers, Pluvialis dominica, and Whimbrels, Numenius phaeopus; Figure 4a,b). While peak biomass is predicted to remain similar for the generic warm site, the model predicted an increase of peak arthropod biomass (80 mg/trap/day, from 30 to 110 mg/trap/day) for the generic cold site (Figure 4a). Such a marked increase in arthropod availability induced by warming would occur during the hatching periods currently observed for most shorebird species breeding in relatively cold Arctic sites (Figure 4a).

4 DISCUSSION

Our results, based on pan-Arctic arthropod monitoring spanning a gradient exceeding 10°C in average daily air temperatures during June and July, provide support to our hypothesis that extended exposure to elevated temperatures can advance arthropod phenology while increasing their potential biomass accessible to consumers like Arctic-nesting shorebirds. Our investigation revealed a strong positive correlation between temperature and arthropod availability (peak or seasonal biomass) below specific temperature thresholds. Moreover, the correlation between temperature and arthropod phenology, as expressed by the date of peak biomass available to birds, was relatively moderate across our large temperature gradient, with an average of a 1-day shift for an increase of 25 DD (roughly equivalent to a 0.6°C difference in average summer temperature). This shift is small when considering the broad spectrum of annual arthropod peak dates encountered by birds breeding at several Arctic sites (intra-site variation reaching >15 days; Kwon et al., 2019; this study). Overall, our findings suggest that prolonged exposure to elevated temperatures can have a positive effect on arctic arthropod availability that may help counteract warming-induced phenological shifts of arthropods and reduce the risk of trophic mismatch for shorebirds.

The non-linear relationship between temperature and biomass of arctic arthropods outlined in our study is a novel finding. The physiological functions of arthropods such as metabolism and growth rates, are largely constrained by temperature, and some arctic arthropods could benefit from a temperature increase in northern regions (Barrio et al., 2017; Bolduc et al., 2013; Shaftel et al., 2021). Although positive responses may be the case for some arthropod groups and during specific timeframes, recent studies have also shown negative or heterogeneous responses of some arctic arthropods to rising temperatures (Bowden et al., 2018; Høye et al., 2021; Loboda et al., 2017). Above a temperature threshold, the positive effects of temperature on biomass and activity rate of some surface-active and low-flying arthropods could thus be counterbalanced by the negative effects on other taxa, leading to the observed null effect of temperature increase on arthropod availability (see also Høye et al., 2021; Koltz et al., 2018).

A wide array of mechanisms can lead to a negative effect of temperature on arctic arthropods (Høye, 2020). For example, higher summer temperatures can lead to a decrease in arthropod body size, possibly due to metabolic costs (Bowden et al., 2015), a reduction of arthropod activity rates above a certain temperature threshold (Asmus et al., 2018) and reduced vegetation nutritive quality (Welti et al., 2020). Warmer winter conditions can also have a negative impact on summer arthropod abundance through a higher frequency of freeze–thaw events or reductions in insulation due to less snow cover during winter (Ávila-Jiménez et al., 2010; Everatt et al., 2015; Høye et al., 2021). Changes in water availability induced by changes in patterns of precipitation, melting of permafrost, or higher evaporation rates, can in turn negatively affect the abundance of some arthropod species (Ávila-Jiménez et al., 2010; Bowden et al., 2018).

After controlling for cumulative thawing degree-days (DD), the relationships we documented between arctic arthropod phenology and other weather variables are generally consistent with previous findings. For example, we found that later snowmelt was associated with later peak date, which is consistent with previous studies that identified snow dynamics as a major predictor of arctic arthropod phenology (Høye, 2020; Høye & Forchhammer, 2008a). On the other hand, Kwon et al. (2019) reported a strong negative effect of timing of snowmelt on peak arthropod biomass available to birds. We did not detect such an effect, likely due to our reliance on less precise estimates of snowmelt date. Finally, summer precipitation is known to negatively affect daily availability of arthropods (Asmus et al., 2018; Shaftel et al., 2021; Tulp & Schekkerman, 2008), but few studies investigated the effect of summer precipitation on arthropod phenology (negligible effects reported in Saalfeld et al., 2019). Use of high quality local weather data could improve the understanding of the positive effect of summer precipitation that we found on arthropod peak date (see Appendix S5).

The performance of space-for-time substitution to predict the response of ecological systems to global warming can be highly variable (Damgaard, 2019). Spatial variation often results from long-term processes that can lead to misestimating short-term responses to warming (Elmendorf et al., 2015). The main assumptions of our approach were that (1) arthropod communities assemblages sampled across our broad temperature gradient would respond to a given temperature increase in a similar way or (2) arthropod assemblages would change relatively quickly following warming, and the new assemblage would respond to a temperature increase in a similar way to assemblages currently found at warmer sites. Although the same dominant taxa are usually present in arctic arthropod assemblages (Bolduc et al., 2013; Shaftel et al., 2021) and rapid climate-induced changes in the composition of these assemblages have been observed (Koltz et al., 2018; Loboda et al., 2017), the validity of these assumptions remains uncertain. Moreover, some ecological processes indirectly affecting the biomass of arthropods available to consumers, such as vegetation changes, may operate over longer timescales. The effect of warming on arthropod availability could be assessed by integrating longer time series data from several sites encompassing diverse environmental conditions and experiencing temperature increases over time (Damgaard, 2019).

Although climate change has led to significant changes in the timing of critical life history events among interacting species (e.g. Schmidt et al., 2023), the prevalence of warming-induced mismatch remains low in terrestrial study systems linking the level of asynchrony to individual fitness (Kharouba & Wolkovich, 2023). Our results also suggest that some generalist insectivorous arctic bird populations may be less vulnerable to mismatch than expected due to a potential warming-induced increase in food availability. Moreover, some shorebirds (and other insectivorous birds) can advance their breeding dates under warmer conditions (Kwon et al., 2019; Liebezeit et al., 2014; Ruthrauff et al., 2021). Although this advancement may not perfectly track phenological shifts in environmental conditions (Saalfeld & Lanctot, 2017), it should also reduce the risk of trophic mismatch. Some arctic shorebird species, like Dunlin (Calidris alpina) and Sanderling (Calidris alba), are already breeding late relative to seasonal peaks in arthropod abundance (McKinnon et al., 2012, 2013; Reneerkens et al., 2016; see also Figure 4) and hence may not benefit from a potential positive effect of temperature on arthropod peak biomass. Birds having more specialized diets or those dependant on highly nutritional food resources could also be more vulnerable to warming-induced changes in prey phenology and quality (Arnold et al., 2010; Wilde et al., 2020; Zhemchuzhnikov et al., 2022). Hence, further investigations may be useful to fully quantify the risk of mismatch for arctic insectivorous birds, while considering that higher temperatures encountered by chicks could provide thermogenic relief that can compensate (or not) for their lack of synchrony (Lameris et al., 2022; McKinnon et al., 2013; Saalfeld et al., 2021).

Climate warming can lead to significant shifts in the timing of key life history events in Arctic ecosystems (Post et al., 2018). Based on a space-for-time substitution, our pan-arctic study indicates that the positive effects of prolonged exposure to elevated temperatures on food availability may help counteract warming-induced phenological shifts in peak food availability for some arctic birds. Incorporating time series data from Arctic sites where temperatures have been increasing over time and the inclusion of weather parameters outside the breeding season could strengthen our findings. Additionally, employing higher arthropod taxonomic resolutions may help pinpoint the specific ecological processes driving warming-induced changes in arthropod availability for birds.

AUTHOR CONTRIBUTIONS

Aurélie Chagnon-Lafortune: Data curation; formal analysis; investigation; methodology; project administration; software; visualization; writing – original draft. Éliane Duchesne: Software; visualization; writing – original draft. Pierre Legagneux: Conceptualization; writing – review and editing. Laura McKinnon: Investigation; methodology; writing – review and editing. Jeroen Reneerkens: Methodology; writing – review and editing. Nicolas Casajus: Software. Kenneth F. Abraham: Investigation; writing – review and editing. Élise Bolduc: Data curation; investigation; methodology. Glen S. Brown: Investigation; writing – review and editing. Stephen C. Brown: Investigation; writing – review and editing. H. River Gates: Investigation; writing – review and editing. Olivier Gilg: Investigation; supervision; writing – review and editing. Marie-Andrée Giroux: Investigation; writing – review and editing. Kirsty Gurney: Investigation; writing – review and editing. Steve Kendall: Investigation; writing – review and editing. Eunbi Kwon: Investigation; writing – review and editing. Richard B. Lanctot: Investigation; writing – review and editing. David B. Lank: Investigation; writing – review and editing. Nicolas Lecomte: Investigation; writing – review and editing. Maria Leung: Investigation; writing – review and editing. Joseph R. Liebezeit: Investigation; writing – review and editing. R. I. Guy Morrison: Investigation; writing – review and editing. Erica Nol: Investigation; writing – review and editing. David C. Payer: Investigation; writing – review and editing. Donald Reid: Investigation; writing – review and editing. Daniel Ruthrauff: Investigation; writing – review and editing. Sarah T. Saalfeld: Investigation; writing – review and editing. Brett K. Sandercock: Investigation; writing – review and editing. Paul A. Smith: Investigation; writing – review and editing. Niels Martin Schmidt: Investigation; writing – review and editing. Ingrid Tulp: Investigation; writing – review and editing. David H. Ward: Investigation; writing – review and editing. Toke T. Høye: Conceptualization; investigation; methodology; supervision; writing – review and editing. Dominique Berteaux: Supervision; writing – review and editing. Joël Bêty: Conceptualization; investigation; methodology; project administration; supervision; writing – original draft.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest.