<p> Size- and condition-dependent predation: a seabird disproportionately targets substandard individual juvenile salmon 1 1 Appendix S2: Electivity index with standardized lengths</p><p>2 To account for any temporal mismatches between survey and colony sampling and the potential for </p><p>3 growth, we recalculated the Electivity Index (ɛ; Chesson 1983) by standardizing all fish lengths to July </p><p>-1 4 1 by assuming an average growth rate of 1mm•d (Groot and Margolis 1991). For example, if a 120 </p><p>5 mm fish was captured July 5, its standardized length would be 116 mm. We assumed condition index </p><p>6 (+/-) was the same as it is not expected to vary that quickly. In fact, controlled starvation experiments </p><p>7 demonstrate that salmon can be food deprived for 3 weeks without seeing any effects on the sign of </p><p>8 length-weight residuals; only after 4 - 6 weeks of starvation is a decline noted (Hertz et al. submitted-</p><p>9 currently in revision).</p><p>10</p><p>11 Fish (trawl and predated) were classified into 4 categories based on fork length and condition. Fish </p><p>12 were either small or large relative to the pooled mean fork length of 122 mm and in good or poor </p><p>13 condition depending on a positive or negative weight residual respectively. The proportion of fish in </p><p>14 each category for each species and year was calculated for trawl-caught and predated fish separately. </p><p>15 Note that the trawl proportions were adjusted by the total catches (all fish were counted) given the </p><p>16 selective sampling of these species (with respect to the maximum number of individuals measured </p><p>17 aboard the research vessel). For each prey category, Electivity Index ɛ contrasts the diet proportions to </p><p>18 the relative abundance as indexed by the trawl survey:</p><p>19</p><p>20 where m = number of potential dietary types (in this case 12: 3 species x 2 size classes (small or large) </p><p>21 x 2 condition levels (good or poor)) and αi = Manly’s Selection Index for prey type i: </p><p>22 Size- and condition-dependent predation: a seabird disproportionately targets substandard individual juvenile salmon 2 23 where ri,rj = proportions of prey types i and j in the diet (i and j= 1, 2, 3,..., m); ni, nj = proportions of </p><p>24 prey types i and j available in coastal waters. Values of ɛ vary from -1 (strong avoidance) to +1 (strong </p><p>25 preference).</p><p>26 Although specific values of ɛ changed slightly from the original, trends were similar as there were no </p><p>27 substantial switches in selectivity/avoidance (Figure S1). In both years, the electivity index was higher </p><p>28 for small salmon of either good or poor condition, with the exception of large pink salmon in poor </p><p>29 condition in 2013. Within a size category, fish in poor condition consistently had the higher electivity </p><p>30 index. Small sockeye salmon in poor condition were strongly selected in both years relative to all other </p><p>31 prey classes and large positive sockeye salmon and chum salmon were very strongly avoided in both </p><p>32 years, while large positive pink salmon were strongly avoided in 2012 and moderately avoided in 2013.</p><p>33 In 2012, small positive sockeye salmon were moderately selected along with small negative pink </p><p>34 salmon. In 2013, large and small negative pink salmon were also selected. In contrast to the original </p><p>35 results (main text Figure 4), small-poor condition chum in 2012 went from weakly selected to weakly </p><p>36 avoided and in 2013 there was increased selection for small-poor condition pink. </p><p>37</p><p>38 References</p><p>39 Chesson, J. 1983. The estimation and analysis of preference and its relationship to foraging models. </p><p>40 Ecology 64:1297-1304.</p><p>41 Groot, C. and Margolis, L. (eds). 1991. Pacific salmon life histories. University of British Columbia </p><p>42 Press, Vancouver. </p><p>43 Hertz, E., M. Trudel, M.K. Cox, and A. Mazumder. Reconsidering the effects of starvation in food web</p><p>44 studies. Ecology and Evolution in revision.</p><p>45 Size- and condition-dependent predation: a seabird disproportionately targets substandard individual juvenile salmon 3 46 Figures</p><p>47 Figure S1: Electivity index (ɛ) for different salmon prey classes using standardized lengths where </p><p>48 SE=sockeye salmon, PK=pink salmon, CM=chum salmon, large >122mm, small<122, + (white bars) is</p><p>49 a positive weight residual and - (black bars) is a negative weight residual. Fish lengths were </p><p>-1 50 standardized to July 1 by assuming an average growth rate of 1mm•d (Groot and Margolis 1991). Size- and condition-dependent predation: a seabird disproportionately targets substandard individual juvenile salmon 4</p><p>SE large + 2012 SE large - SE small + SE small - PK large + PK large - PK small + PK small - CM large + CM large - CM small + CM small -</p><p>SE large + 2013 SE large - SE small + SE small - PK large + PK large - PK small + PK small - CM large + CM large - CM small + CM small -</p><p>-1.00 -0.50 0.00 0.50 1.00</p><p>51</p><p>52</p><p>53 Figure S1</p>