Les δ¹³C varient de −26,5‰ pour les plantes dites en C₃ à −12,5‰ pour celles en C₄. Par rapport aux plantes en C₃, la photosynthèse de type C₄ présente un meilleur rendement pour les basses pressions de CO₂ [6]. La teneur en CO₂ de l'atmosphère est donc un facteur au moins aussi important que l'humidité ou la température pour le développement des plantes en C₄ [6,22].

Le δ¹³C des coquilles des autruches actuelles présentent un enrichissement de 16,2‰ par rapport à celui de l'alimentation [34]. Pour les fossiles, nous considérons que leur physiologie n'a pas changé depuis le Miocène. Le δ¹³C de leur nourriture est fonction des proportions consommées de plantes en C₃ et en C₄ [13]. En tenant compte de la variabilité (±3‰) des δ¹³C de chaque type de plantes (lié à l'aridité, aux précipitations [7,9–11]), on peut donc calculer le pourcentage de chaque type consommé par individu : $[\%\mathrm{de}\mathrm{plantes}\mathrm{en}{\mathrm{C}}_3]=[(-\delta {}^{13}{\mathrm{C}}_{\mathrm{coq}}+3,7)/14]\times 100$ et $[\%\mathrm{de}\mathrm{plantes}\mathrm{en}{\mathrm{C}}_4]=[1-(-\delta {}^{13}{\mathrm{C}}_{\mathrm{coq}}+3,7)/14]\times 100$. Ces relations ont été vérifiées en élevages à nourriture contrôlée [34] ; toutefois, dans leur milieu naturel, ces animaux présentent une préférence pour les plantes en C₃ [12], qui introduit un biais dans la reconstitution des paléoenvironnements.

Les fluctuations du δ¹³C (Tableau 1 et Fig. 2) ont été établies pour les 20 derniers millions d'années (l'âge des coquilles de type Aepyornithoı̈de d'Elisabethfeld est estimé à 19–20 Ma par les mammifères). Le δ¹³C moyen est de −8,6‰ au Miocène inférieur. Au Miocène moyen, il augmente d'abord légèrement (−8,25‰ dans la biozone à Namornis oshanai, 15–16 Ma), puis chute en deux étapes : (1) vers 15 Ma, il diminue de −1,31‰ entre les biozones à N. oshanai et à D. corbetti, où le δ¹³C moyen est alors de −9,56‰, puis reste ensuite stable (−9,51‰) durant la biozone à D. spaggiarii (12–14 Ma) ; (2) vers 12 Ma, une nouvelle diminution de −1,25‰ survient entre les biozones à D. spaggiarii (12–14 Ma) et à D. wardi (10–12 Ma), où le δ¹³C moyen atteint −10,76‰. Durant le début du Miocène supérieur, le δ¹³C se stabilise (−10,89‰ dans la biozone à D. laini, 8–10 Ma), puis diminue légèrement, pour atteindre −11,26‰ dans la biozone à S. karingarabensis (Miocène supérieur–Pliocène basal, 8–5 Ma). Durant la biozone à S. daberasensis (2–5 Ma), le δ¹³C remonte de près de 1‰, pour atteindre une valeur de −10,23‰ au cours du Pliocène. Enfin, les oeufs actuels (S. camelus) présentent un δ¹³C de −5,62‰.

5.1 Des plantes en C₄ présentes en Namibie dès le Miocène inférieur ?

5.2 Conséquences paléoclimatiques

L'évolution du δ¹³C des coquilles met en évidence une modification du régime alimentaire des ratites, indiquant une variation du pourcentage relatif des plantes en C₃ et en C₄ dans l'écosystème du Namib. Les faibles concentrations en CO₂ atmosphérique au Miocène inférieur semblent être le paramètre majeur expliquant la présence de plantes C₄ dans le milieu. Durant le Miocène moyen et terminal, les conditions régionales de température et de précipitations exercent ensuite un effet de plus en plus important sur la composition de l'écosystème namibien.

The evolution of eggshell δ¹³C could be interpreted in term of C₃/C₄ diet evolution from Eq. (3). Taking into account the δ¹³C variability of plant (±3‰), the diet of Early and Middle Miocene species (20–14 Ma), is made of 65 to 100% of C₃ plants (most probable value 85%). The diminution of the δ¹³C occurring between 14 and 12 Ma leads to a 75 to 100% contribution of C₃ plants (probable value: 95%). C₃ plants reach 85 to 100% between 12 and 5 Ma (probable value: 100%). For modern Namibian ratites, a large proportion of C₄ plants should be present in the food diet to explain the eggshell δ¹³C. C₃ plants represent only between 45 and 85% (probable value 65%) of the diet. This estimation is in accordance with the modern Namibian ecosystem composition (30% of C₃ plants [26]), in relation with the ostriches' dietary preference for C₃ plants [12].

Numerous studies have shown that environmental parameters can influence the value of the δ¹³C of plants: aridity increases the δ¹³C value in C₃ plants (0.33‰ per 100 mm decrease in precipitation, [32]). An increase of 100 ppm in atmospheric CO₂ concentrations leads to a decrease of 2‰ in plant δ¹³C; the δ¹³C of atmospheric CO₂ has also an impact [9]. In spite of these relationships, we should test if δ¹³C fluctuations in the Upper Tertiary could be explained by climatic variations without diet composition fluctuations. For this, we can use the equation proposed by Hatté and Antoine [11]:

Pre-industrial period parameters could be [CO₂]₀=280 ppmv, $\delta {}^{13}{\mathrm{C}}_{\mathrm{C}3\mathrm{plant}0}=-26$‰, $\delta {}^{13}\mathrm{C}{\left[{\mathrm{CO}}_2\right]}_0=-6.5$‰ and Precip₀=50 mm (mean value for the Namib desert). During the Neogene, in the absence of data, we have used the δ¹³C of pre-industrial period as δ¹³C[CO₂]. The Miocene atmospheric CO₂ concentrations during the Miocene could be estimated on the basis of Pagani data [16]. From the beginning of Early Miocene (24 Ma) to Middle Miocene (16–15 Ma), the concentration has first diminished by 110 ppmv, before increasing after 15–14 Ma (220 ppmv, Fig. 2). Related to the installation of the East Antarctic Ice Sheet (as recorded by an excursion of +1‰ in planctonic foraminifera δ¹⁸O at about 14 Ma), this increase continues until 10 Ma (280 ppmv). Nowadays, in Namibia, annual precipitation means are comprised between 150–400 mm in grassland savannah regions and between 100–150 mm in semi-arid regions; they are lower than 50 mm in arid areas.

From Eq. (4), we can calculate a theoretical eggshell δ¹³C for different rates of precipitations (Table 2). For modern Namibian species, the use of Eq. (4) (Table 2) confirms that with the actual rate of precipitation (<50 mm), the eggshell δ¹³C could not correspond to a 100% C₃ diet. A participation of C₄ plant of the order of 40% is needed (see above).

Results are compatible with a 100% C₃ diet during the Upper and Middle Miocene, if the precipitation were lesser than 50 mm in the first case and comprised between 50 and 100 mm in the second one. These precipitation estimations agree with sedimentological and palaeontological data.

Studies from Elisabethfeld [19] and Arrisdrift [20] show that, during Early Miocene, the climate was more humid and warmer than it is today. During Middle Miocene, an important climatic change occurred in the southern hemisphere. The installation of the East Antarctic Ice Sheet comes along with important modifications in oceanic circulation patterns [34]. In Namibia, these changes correspond to the development of arid conditions [33], as indicated by the first occurrences of aeolianites during the Namornis oshanai biozone at Awasib. This aridification of the southwestern Africa coastal zone confirms the passage from a warm, humid weather to a drier climate during the Early to Middle Miocene transition.

For the Early Miocene species, the eggshells δ¹³C values are only compatible with a diet composed of 100% of C₃ plants, if the precipitations are lower than 50 mm. This is incompatible with the palaeontological and sedimentological data, which indicate a wetter climate during this period. For precipitations in the range 200–250 mm, the theoretical δ¹³C (Eq. (4)) is similar to the measured δ¹³C, only if we admit a large participation of C₄ plants in the diet of these birds. This value (about 40%) is undoubtedly too high, but proves a presence of C₄ plants in the Early Miocene ecosystem. This development of C₄ plants could be related to the weak atmospheric CO₂ concentrations (Fig. 2). Previous studies have documented C₄ plant occurrences during the Early–Middle Miocene in some peculiar environments. Their presence is recorded in the δ¹³C of Rhinocerotidae and Proboscidea teeth from the Tugen Hills (Kenya) at 15.3 Ma [15]. The site of Fort Ternan (14 Ma) has also yielded C₄ Graminae (Chloridoidae), indicating the existence of open forest ecosystems, with a dominance of grasses [25]. The rarefaction of C₄ plants in the ecosystem during Middle and Upper Miocene could reflect an increase in the atmospheric CO₂ concentration.

The eggshell δ¹³C increases by 1‰ during the Late Miocene and Pliocene and by 4.5‰ from the Pliocene to the present day (Fig. 2). Despite the ostriches' preference for C₃ plants [12], the modern Namibian δ¹³C eggshell corresponds to a diet comprising only 65% C₃ plants (cf. above). In contrast, the δ¹³C value suggests a greater presence of this plant type in the diet during the Late Miocene. Namib data confirms, therefore, that apart from changes related to regional climatic fluctuations, the Pliocene was a crucial period in the development and ecology of C₄ plants in the continental biosphere. It was at this epoch that they became preponderant in some ecosystems [3,15,23,24].

After important fluctuations during the Holocene [14], the Modern (320 ppmv) and Late Miocene (280 ppmv), atmospheric CO₂ concentrations are now close together. From Late Miocene to Present, the CO₂ concentrations do not seem to be the main control of the C₃/C₄ plant ratio in the Namibian desert ecosystem. Abundant roots and incipient palaeosoils in Awasib aeolianites show that Upper Miocene was a less arid period than Present. The temperature and rainfall patterns seem to control the increase of the C₄ plants in the Namib Desert ecosystem between Late Miocene and today.