1 Trait selection during food web assembly: 2 The roles of interactions and temperature 3 Online Appendices 4 5 Isabelle Gounand*1,2,3,4,5, Sonia Kéfi1, Nicolas Mouquet1 and Dominique Gravel2,3 6 7 Affiliations 8 1. Institut des Sciences de lβEvolution, Université de Montpellier II, CNRS, IRD, CC 065, 9 Place Eugène Bataillon, 34095 Montpellier Cedex 05, France 10 2. Université du Québec à Rimouski, Département de Biologie, Chimie et Géographie, 11 300 Allée des Ursulines, Québec G5L 3A1 Canada. 12 3. Quebec Center for Biodiversity Science. 13 4. Department of Evolutionary Biology and Environmental Studies, University of Zurich, 14 Winterthurerstrasse 190, CH-8057 Zürich, Switzerland. 15 5. Eawag, Swiss Federal Institute of Aquatic Science and Technology, Department of 16 Aquatic Ecology, Überlandstrasse 133, CH-8600 Dübendorf, Switzerland. 17 18 19 *Corresponding author: [email protected] 20 List of Supplementary Material 21 Online Appendix A: Equilibria of small modules (Table 3) 22 Online Appendix B: Model comparison (Table 4) 23 Online Appendix C: Interaction domain (Fig. 6) 24 Online Appendix D: Invasion analyses, and the effect of temperature 25 1) Invasion criteria (Table 5 and Figs 7 and 8) 26 2) Effect of temperature (Table 6 and Fig. 9) 27 2 28 29 30 31 Appendix A 32 Equilibria of small modules 33 Table 3 34 Equilibria for models 1 to 4 when one producer P (3.1), or one herbivore H feeding on 35 one producer P (3.2), are present (small modules). For clarity we used the following πΈ(πβπ0 ) ππ¦ π ππβ0.25 π πππ0 36 symbols: ππ = 37 the metabolic rate, and ππ = 38 intractable for model 4 when herbivore is present (see Table 1 for symbols). 3.1 Without ππ ππ₯ = π π , the ratio of the biological rates. Equilibria were π¦π Model R* P* Model 1 ππ ππ ππ πΌππ ππ’π‘π β ππ ππ Model 2 ππ ππ 1 πΌππ ( β ππ’π‘π ) ππ ππ Model 3 ππ ππ πΌ ( β ππ’π‘π ) ππ ππ Model 4 ππ ππ’π‘π (1 + ) ππ ππ 1 πΌππ ( β ππ’π‘π ) ππ π (1 + ππ’π‘π ) π ππ πΌ ππ» ππ» ππ» 1 (π π π β β ππ ) ππ» π π ππ» ππ» 1 (π π π β β ππ ) ππ» π π ππ» ππ» (π π β β ππ ) ππ» π Herbivore (Fig. 1a) ππ the maximum consumption rate, ππ = πΈ(πβπ0 ) ππ₯ π ππβ0.25 π πππ0 Model 1 ππ’π‘π + 3.2 With Herbivore ππ ππ» ππ» ππ» πΌ Model 2 (Fig. 1b and 1c) ππ ππ’π‘π + π π» ππ» πΌ Model 3 ππ ππ’π‘π + π π» ππ H* 39 3 40 Appendix B 41 Model comparisons 42 Table 4 43 Model comparisons for a compartment i fed upon by a compartment j, with π΅π the density 44 of compartment π. For clarity we used the following symbols: ππ = ππ¦ π ππβ0.25 π 45 maximum per capita uptake rate, and ππ = ππ₯ π ππβ0.25 π 46 allometric constants ππ₯ π and ππ¦ π (for the biological rates measured at temperature T0) 47 are different between producers and herbivores but they do not vary within a trophic 48 level in the multispecies simulations. Models 3 and 4 are variants of model 2 and the 49 differences from model 2 are highlighted with grey background. The supply and loss rate 50 of the inorganic resources are the same for all models: πΌ β ππ’π‘π π π . (see Table 1 for 51 symbols) πΈ(πβπ0 ) πππ0 πΈ(πβπ0 ) πππ0 the the metabolic rate. Note that Description Uptake Production Catabolism Exportation Model 1 Classical ConsumerResource model βππ π΅π π΅π +ππ ππ π΅π π΅π βππ π΅π - βππ π΅π π΅π +ππ ππ΅ π΅π π΅π βππ π΅π - βππ π΅π π΅π ππ +ππ π΅π π΅π βππ π΅π - βππ π΅π π΅π +ππ ππ΅ π΅π π΅π βππ π΅π βππ’π‘π π΅π Bioenergetics Model Model 2 Efficiency in production Model 3 Efficiency in uptake Model 4 Supplementary losses 52 53 54 55 4 56 57 APPENDIX C 58 Interaction domain 59 60 Fig. 6 61 Interactions between herbivores and producers are based on their body mass π. We 62 defined the domain of interaction (grey area) using the 10th and 90th quantile regressions 63 on an empirical relationship between the logarithm of body mass of preys and predators 64 (Brose et al. 2006a, Gravel et al. 2013). Given its own body mass ππ» , an herbivore π» 65 feed on all the producers whose log(π) fall into the domain of interactions (dotted lines). 66 Given our initial distributions of body mass species (logβ‘(π) β π°[10β8 , 108 ]), the largest 67 producers escape herbivory 68 5 69 APPENDIX D 70 Invasion analysis, and the effect of temperature 71 We consider the situations where a producer or an herbivore tries to invade an 72 ecosystem as described in the Fig. 1a, 1b and 1c. 73 1) To invade the ecosystem, a species must grow when initially rare despite the 74 competition with the residents. 75 2) Changes in temperature may affect this outcome through variation in resource 76 availability or in species performance.. 77 78 1) Invasion criteria 79 When a competitor is already present in the ecosystem, the invader must win the 80 competition to persist. It means that its growth rate must be positive when it tries to 81 invade the ecosystem at equilibrium. For instance in the case described in Fig. 2b, we 82 ask whether the producer P2, initially rare (π2 β 0), and sharing the same resource than 83 P1 and the same herbivore H, can invade the system and exclude P1. The criteria to be 84 satisfied is: 85 86 87 ππ2 | > 0β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘[13] ππ‘ π β,π1β,π» β,π2βͺπ1β β‘ Taking model 1 as an example, the system can be written: ππ = πΌ β ππ’π‘π π β (ππ1 π1 + ππ π·π )π ππ‘ ππ1 = ππ1 ππ1 π1 π β ππ1 π1 β ππ» π1 π» ππ‘ β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘[14] ππ2 = ππ2 ππ2 π2 π β ππ2 π2 β ππ» π2 π» ππ‘ ππ» { ππ‘ = (ππ1 π1 + ππ π·π )ππ» ππ» π» β ππ» π» 6 88 Since π2 tends to 0 at invasion, we can approximate the system by deleting the terms 89 containing π2 in the equation of R and H in the system [14] (in bold and red). Then the 90 invasion criteria becomes: ππ2 | = ππ2 ππ2 π2 π β β ππ2 π2 β ππ» π2 π» β > 0β‘β‘β‘β‘β‘β‘β‘β‘β‘[15] ππ‘ π β,π1β,π» β,π2 βͺπ1β 91 πΌ 1 92 with: π β = 93 The condition [15] can be written as in Table 5. At the upper trophic level, this condition 94 implies the minimization of the resource requirement (R* rule). For the producers under 95 herbivory, the apparent competition changes this condition: the winner will be the one 96 that sustains the highest densities of herbivores at equilibrium. Table 5 shows the 97 invasion condition for the competition in the four models derived from the same 98 procedure. 99 Our interpretation of trait selection is as follows: ππ ππ» ππ’π‘π + 1 ππ» ππ» and π» β = π (ππ1 ππ1 π β β ππ1 ) (see Appendix A for the equilibria) π» 100 For the models 1 and 2, the invasion criteria involves combinations of several traits in 101 which traits of P1 cannot be expressed relative to traits of P2. We derived the direction of 102 trait selection from the form of the invasion condition and from supplementary numerical 103 analyses (Figs. 7 and 8): 104 For model 1, since the Rβ is positive by definition, the invasion condition (Table 5) 105 is more easily satisfied if ππ2 ππ2 is large relative to ππ1 ππ1 (Fig. 7) and if ππ2 is small 106 compared to ππ1 . We conclude that competition should select species with higher 107 conversion efficiency and attack rate, and lower mortality rate. 108 109 For model 2, the invasion criteria (Table 5) is more easily satisfied if ππ2 is large relative to ππ1 and ππ2 small compared to ππ1 . The invader P2 is more likely to 7 110 outcompete the resident P1 if it has the smallest body mass possible and the highest 111 conversion efficiency (Fig. 8a compared to Fig. 8b). We conclude that competition 112 should select species with smaller body mass and higher conversion efficiency. 113 For models 3 and 4, the species with the smallest body mass wins the competition 114 (Table 5). 115 8 116 117 Table 5 118 Invasion conditions for competition in small modules (see Figs 1a, 1b, 1c). For clarity, 119 we used the following simplifications: π π = 120 the temperature-dependent term: π(π) = π ππ’π‘π ππ₯π ππ» , π π = π π» ππ» πΈ(πβπ0 ) πππ0 , π π = ππ’π‘π π π¦π , π π = π ππ₯π» π» ππ¦π ππ¦π» , and of . Herbivore Model Producer invasion condition invasion (P2 the invader) condition (H2 the invader) General π 2β < π 1β condition Model 1 ππ2 ππ1 < ππ2 ππ2 ππ1 ππ1 Herbivore Model 2 ππ2 > ππ1 (Fig. 1a) Model 3 ø Model 4 βπ(π) 1 + π π ππ0.25 βπ(π) 1 + π π ππ0.25 2 1 < ππ2 ππ1 5.1 Without General π»1β < π»2β condition 5.2 With ππ2 ππ2 β ππ1 ππ1 > Model 1 Herbivore π2β < π1β ππ2 β ππ1 ππ»2 ππ»1 < ππ»2 ππ»2 ππ»1 ππ»1 πΌ ππ’π‘π + ππ1 π π (Figs 1b and1c) Model 2 1 ππ0.25 2 ( Model 3 ππ2 πΌ β π π₯π π π(π) π π + π 0.25 ππ1 > 1 ππ0.25 1 ) ππ2 < ππ1 ( ππ1 πΌ β π π₯π π π(π) π π + π 0.25 ππ1 ππ»2 > ππ»1 ) ø 121 122 9 123 124 Fig. 7 125 Outcomes of invasions with model 1 (Classical consumer-resource model) for the case 126 shown in Fig. 2b (a producer P2 try to invade an ecosystem at equilibrium where a 127 producer P1 consume a resource R and is fed upon by an Herbivore H), according to the 128 attack rates and conversion efficiencies of P1 and P2. Parameter values: ππ» = 0.8, 129 ππ1 = ππ2 = 0.2; panel a: ππ1 = 0.6 and ππ2 = 0.8; panel b: ππ1 = ππ2 = 0.8; panel c 130 ππ1 = 0.8 and ππ2 = 0.6. See Table 1 for other parameters 131 10 132 133 134 Fig. 8 135 Outcomes of invasions with model 2 (bioenergetic model with efficiency at the 136 production), for the case shown in Fig. 2b (a producer P2 try to invade an ecosystem at 137 equilibrium where a producer P1 consume a resource R and is fed upon by an Herbivore 138 H), according to log10 of body masses and conversion efficiencies of P1 and P2. 139 Parameter values: ππ» = 0.8; panel a ππ1 = 0.2 and ππ2 = 0.8; panel b: ππ1 = ππ2 = 0.8; 140 panel c ππ1 = 0.8 and ππ2 = 0.2. See Table 1 for the other parameter 141 11 142 2) Effect of temperature 143 An increase in temperature has two effects on invasion success. First, it reduces some 144 equilibrium densities. Table 6 shows these changes in equilibrium densities for models 2 145 to 4. In model 2 the main consequence is that the reduction of πβ in absence of 146 herbivores makes harder the success of invasion by herbivores, because of reduced 147 resource availability (P* without the herbivore diminishes). 148 Second, it may affect the outcome of competition. For model 2, the temperature has no 149 effect on the selection at the upper trophic level, but it affects the invasion criteria for the 150 producer under herbivory (see expressions enlightened in red in Table 5). Below we 151 give the extensive analysis of temperature effect on this condition. The results are 152 summarized at the end in italic. 153 In case of herbivory, the condition for the invader P2 (case of Fig. 2b) to successfully 154 invade the ecosystem is: 155 156 πΌ(ππ2 ππβ0.25 β ππ1 ππβ0.25 ) 2 1 π π + π π ππβ0.25 π(π) 1 with the following simplifications: π π = β ππ₯π (ππβ0.25 β ππβ0.25 ) > 0β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘[16] 2 1 ππ’π‘π ππ¦π , π π = π ππ₯π» π» ππ¦π» , and π(π) = π πΈ(πβπ0 ) πππ0 157 158 This condition is valid only if the herbivore has enough resource to invade. This means 159 that the P* without the herbivore must be higher than the expected P* with the herbivore, 160 which can be expressed as follows: 161 1 ππ¦π β‘ππβ0.25 π 162 πΌππ ππ» ( β ππ’π‘ ) > β‘ > 0β‘β‘β‘β‘β‘β‘[17] π πΈ(πβπ0 ) π ππ» π πππ0 π π₯π with the following simplifications: ππ = π π¦π ππ₯π» and ππ» = π π¦π» 12 163 We calculated the partial derivative of the left term of condition [16] with respect to 164 temperature, yielding: βπππ 165 (π + πΈ(πβπ0 ) πππ0 πΈ ππ 2 β‘β‘β‘β‘β‘β‘[18] πΈ(πβπ0 ) 2 ππ πππ0 ) ππ’π‘π ππ₯π» 166 with the following simplifications: π = πΌ(ππ2 ππβ0.25 β ππ1 ππβ0.25 ), π = 2 1 167 π = ππ₯π (ππβ0.25 β ππβ0.25 ) 2 1 168 Since the sign of the partial derivative [18] does not depend on temperature but only on 169 π, the left term of inequality [16] for the successful invasion of P2 is monotonous. The 170 outcome of the competition change with temperature only if the sign of this left term 171 changes with temperature and the condition of herbivore maintenance [17] is still valid 172 for the parameters of P1 and P2. Actually, if the herbivore dies, then the condition for 173 invasion becomes the R* rule (Table 5.1). We derived analytically the critical 174 temperatures for which these three condition changes, ππππ£π2 , ππ»ππππ1 and ππ»ππππ2 , 175 respectively: 176 177 ππππ£π2 = π π¦π ,π=π 1 π» ππ¦π» ππβ0.25 , 1 β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘[19] πΌ(ππ2 ππβ0.25 β ππ1 ππβ0.25 ) ππ’π‘π ππ» ππ¦π» 1 π 2 1 β πππ (( β ) π0 πΈ ππ¦π ππ₯π» ππβ0.25 ) ππ₯π (ππβ0.25 β ππβ0.25 ) 1 2 1 ππ»ππππ1 = 1 πΌππ1 ππ’π‘π ππ» ππ¦π» 1 π β πππ (( β ) π0 πΈ π π₯π ππ¦π ππ₯π» ππβ0.25 ) β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘[20] 1 178 ππ»ππππ2 = 1 πΌππ2 ππ’π‘π ππ» ππ¦π» 1 π β πππ (( β ) π0 πΈ π π₯π ππ¦π ππ₯π» ππβ0.25 ) β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘[21] 2 13 179 Given that the left terms of conditions [16] and [17] are monotonous with temperature, 180 the outcome of competition between P1 and P2 changes with temperature only if: (ππππ£π2 < ππ»ππππ1 ) β© (ππππ£π2 < ππ»ππππ2 )β‘β‘β‘β‘β‘[22] 181 182 183 184 185 186 From [19], [20] and [21], the conditions [22] can be re-written: (0 < ππ2 ππβ0.25 β ππ1 ππβ0.25 2 1 ππβ0.25 β ππβ0.25 2 1 < ππ1 β‘) β© (0 < ππ2 ππβ0.25 β ππ1 ππβ0.25 2 1 ππβ0.25 β ππβ0.25 2 1 < ππ2 )β‘β‘β‘β‘[23] Two cases have to be considered: (a) the derivative [18] is negative, that is: ππ2 ππβ0.25 β ππ1 ππβ0.25 > 0β‘β‘β‘β‘β‘β‘β‘[24] 2 1 187 It follows that the left term of inequality [16] is positive at low temperature and 188 decreases when temperature increases. This means that P2 is more competitive than 189 P1 and invades the ecosystem at low temperature, but its relative competitiveness 190 decreases with temperature (Fig. 9a). It might even fail to invade at high temperature 191 if the following conditions, derived from [23] and [24] are met (Fig. 9b): 192 (ππ2 ππβ0.25 β ππ1 ππβ0.25 > 0β‘) β© (ππ2 < ππ1 ) β© (ππ2 < ππ1 )β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘[25] 2 1 193 194 195 (b) the derivative [18] is positive, that is: ππ2 ππβ0.25 β ππ1 ππβ0.25 < 0β‘β‘β‘β‘β‘β‘β‘[26] 2 1 196 It follows that the left term of inequality [16] is negative at low temperature and 197 increases when temperature increases. This means that P2 is less competitive than 198 P1 and fail to invade at low temperature, but its relative competitiveness increases 199 with temperature (Fig. 9c). It might even succeed to invade at high temperature if the 200 following conditions, derived from [23] and [26] are met (Fig. 9d): 14 (ππ2 ππβ0.25 β ππ1 ππβ0.25 < 0β‘) β© (ππ2 > ππ1 ) β© (ππ2 > ππ1 )β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘β‘[27] 2 1 201 202 203 In summary, in general, the producer with the smallest body mass wins the competition. 204 However, under herbivory, the competitiveness of the smallest producer tends to 205 decrease with temperature (Fig. 9). In the case it has lower conversion efficiency than its 206 competitor, it can even lose the competition if its body mass is not sufficiently smaller 207 than the one of its competitor. In terms of trait selection, this means that the smallest 208 body masses and the highest conversion efficiencies should still be selected but 209 temperature increase may temper the strength of this selection when species traits are 210 very similar. 211 15 212 Table 6 213 Arrows show the direction of changes in equilibrium densities with increasing 214 temperature derived from the examination of equilibria shown in Table 3. Equilibria were 215 intractable for model 4 with the herbivore. Structure Models πΉβ π·β 6.1 Without Model 2 β πβ β Model 3 β πβ β Model 4 π β β non-linear Model 2 π β β β π»β β Model 3 π β β β π»β β Herbivore π―β (Fig. 1a) 6.2 With Herbivore (Fig. 1b and 1c) 216 217 218 219 16 Competition outcome changes 0.15 b 0.10 8 no H MP2 < MP1 0 0.00 2 0.05 4 6 P2 invades 10 20 30 T2 40 0 0 β4 β2 T1 P2 excluded c 0 10 20 30 40 0 10 20 β0 . 1 5 β0 . 1 0 β0 . 0 5 0.00 β2 T2 β6 Invasion condition (if positive, P2 invades) 10 a 0.20 No change 30 40 T1 MP2 > MP1 d 0 10 20 30 40 Temperature (°C) 220 221 Fig. 9 222 Outcomes of competition between P1 and P2 for the case shown in Fig. 2b (with 223 herbivory) with model 2, according to temperature and traits of P1 and P2. The line 224 shows the left part of inequation [16]. In the top panels conversion efficiencies are ππ2 = 225 0.90 < ππ1 = 0.91, and P2 has a smaller body mass than P1: (a) ππ2 = 0.0001 < ππ1 = 226 0.001; (b) ππ2 = 0.0001 < ππ1 = 0.00011. In the bottom panels ππ2 = 0.91 > ππ1 = 0.90, 227 and P2 has a larger body mass than P1: (c) ππ2 = 0.001 > ππ1 = 0.0001; (d) ππ2 = 228 0.00011 > ππ1 = 0.0001. If the difference between ππ1 and ππ2 is small enough, 229 temperature may change the outcome of competition (right panels). In all panels: ππ» = 230 0.5 and ππ» = 0.1; see Table 1 for the other parameters 17
© Copyright 2026 Paperzz