The evolutionary history evident in grass pollen morphology

Abstract

Distinguishing among the species, genera, subfamilies, or tribes of Poaceae would provide significant insight into the evolution and establishment of grasses and grassland communities. Even discrimination at the rank of morphotype would allow a more detailed assessment of changes in grass diversity through time than is possible with traditional palynological methods. One proposed approach to delimiting grass pollen morphotypes focuses on the morphology of the pollen surface ornamentation visible under very high magnification. Introduced a decade ago by Mander et al. (2013), we characterized the complexity of the grass pollen exine using the topological method of subgraph centrality, which describes the relationship of a node (or in this case, a pixel) to all other nodes (or pixels) (Estrada & Rodriguez-Velazquez, 2005). ‘Centrality’ refers to the relative importance of a node to local and global connectivity. So, given a cropped image of the pollen grain exine, binarized into a white foreground and a black background, the connection between pixels and their immediate neighbors can be weighted, with connections within the foreground or the background holding high weight and connections across the two assigned as low weight. Expanding the network outward, pixels that are more centrally connected within a defined area will hold a higher value than those at a boundary (Fig. 2). This provides a dynamic range of values that quantitatively characterize complex shapes.

Wei et al. are the first to apply Mander et al. (2013)'s method to the fossil record. They focus on the shape of the pollen surface foreground visible in scanning electron micrographs. This foreground is defined by the top two (SC2) or top three (SC3) quantiles of pixel brightness in these grayscale images. Pixels from these binarized images are ranked by their subgraph centrality values and the number of foreground objects defined at 5% threshold increments is tallied. This produces two vectors, each with 19 elements, that together with the size and density of ornamentation, describe the morphological complexity of a pollen grain's surface. Wei et al. use these morphological characters to define a morphospace for grass pollen. A morphospace captures the theoretical hyperspace of all possible variations in an organism's form (McGhee, 2006), and the changes in the area occupied within a morphospace represent evolutionary shifts in a taxon's morphology.

Wei et al. show that grass pollen morphology, already simple and psilate, has further simplified from the early Miocene, with surface ornamentation getting progressively less and less complex. This is the first demonstration of morphological evolution in grass pollen on evolutionary timescales. Notably, the morphospace occupied by the pollen of living Neotropical grass species in the study is sparsely occupied by the fossil specimens. There is limited overlap between the pollen grain morphology of extant and fossil species, as well as between early Miocene and late Miocene specimens. This points to the potential extinction or extirpation of many Neotropical fossil morphotypes. The morphospace progression of grass pollen morphology from past to present suggests either morphological drift with taxonomic turnover or directed selection for simpler exine ornamentation in Neotropical grasses over the last 23 million years.

Wei et al. rule out the fossilization process and abiotic factors as potential drivers of this shift and suggest that wind pollination may play a role in selection for a simpler exine. But in a recent study on Quaternary grass pollen (Adaïmé et al., 2024b), we offer an alternative hypothesis. The exine of grass pollen may reflect the evolution of different resource allocation strategies. Grass species mostly employ one of two photosynthetic pathways (C₃ or C₄), which have consequences for carbon allocation within the plant. C₃ grasses have leaves with higher tissue density, while C₄ grasses invest more in their root systems (Atkinson et al., 2016). This trade-off in carbon allocation between the shoot and root systems may be reflected in not only the leaves and roots, but also flowers, seeds, and pollen. We observed thinner walls and simpler ornamentation in the pollen of extant C₄ grasses (Adaïmé et al., 2024b), and a similar pattern can be observed in the results from Wei et al. When living species are color-coded by photosynthetic pathway, it becomes clear that while C₃ and C₄ grass pollen overlap morphologically, the fossil specimens overlap the morphospace occupied by C₃ more than that of C₄ species (Fig. 3). C₃ grass pollen is characterized by higher PC1 scores than C₄ grasses and older fossil specimens are characterized by higher PC1 scores than younger fossil specimens. This suggests that the first principal component of the morphological measurements from Wei et al. captures a shift in photosynthetic pathway and perhaps a shift in carbon allocation between roots and shoots, similar to that observed in Adaïmé et al. (2024b). This shift is dramatic but expected, given the global expansion of C₄ grasslands in the late Miocene (Osborne, 2008) likely resulting from increased fire frequency (Hoetzel et al., 2013) and the drawdown of atmospheric CO₂ (Brown et al., 2022).

The results from Wei et al. and other recent studies (Romero et al., 2020a; Adaïmé et al., 2024a,b) clearly demonstrate how much we have underestimated the extent of ecological and evolutionary information preserved in pollen morphology, particularly within the fine-scale structures evident under high magnification. However, the extended delay between the publication of the subgraph centrality method by Mander et al. (2013) and its application by Wei et al. speaks to the broader challenge of gathering the image datasets needed for these intensive morphometric analyses. Producing 1157 scanning electron micrographs is no small feat. That 354 of those were of fossil specimens is even more remarkable. Few researchers have attempted to gather an electron micrograph library of this scale, but Wei et al. demonstrate the discoveries that are possible with intensive microscopic imaging.

Electron microscopy, alongside comparable methods like optical superresolution (Sivaguru et al., 2018; Romero et al., 2020b; Fig. 1), hold a critical place in pollen analysis. Arguably, high-resolution imaging and image analysis should play a larger role in our field. The morphology we are able to observe influences both the hypotheses we test and the robustness of our results (Mander & Punyasena, 2014). Large-scale microscopic imaging of living and fossil pollen and spores and the quantification of the traditional qualitative terminology of pollen analysis is the next frontier of palynology. It is the critical first step in decoding the evolutionary information contained within the microscopic morphology of pollen and spores.

The New Phytologist Foundation remains neutral with regard to jurisdictional claims in maps and in any institutional affiliations.