Tree height varies across environments, with taller individuals found in cool, moist habitats and shorter trees in drier regions. Within species, trees can exhibit height variation due to environmental factors such as drought-induced dieback. A key question is what drives changes in leaf structure with increasing height–whether some trait values cannot be produced under the developmental conditions at treetops or whether differences arise because natural selection favors particular trait values at different canopy positions. Some hypotheses suggest that increasing height imposes limits on mature leaf traits, making some structural changes developmentally inevitable. However, selection could also favor structural changes within wide fields of developmentally possible trait configurations. We examined leaf epidermal cell size distributions in Bursera simaruba and Eucalyptus camaldulensis from seedlings to maximum tree heights in situations in which seedlings to adults were all exposed to full sun and thus had all “sun” leaves. We found that in general, cell sizes increased, variance remained high, and distributions did not systematically shift with height. These results indicate that, rather than reflecting a developmental inability to produce certain leaf epidermal cell sizes at greater heights, the patterns we observed are better explained by selection simply favoring some cell sizes from among the many that development can produce.

The turgor limitation hypothesis suggests that if a tree is approaching or has reached its height limit under given environmental conditions, its leaf cells should display signs of restricted expansion due to reduced turgor at greater heights. This hypothesis implies that these conditions "limit" the tree from producing the cell sizes that natural selection would otherwise favor. As a result, natural selection should favor individuals capable of producing cell sizes as close as possible to the optimal values. The turgor limitation hypothesis thus predicts a distinct cell size frequency distribution, characterized by strong negative skewness and low variance.

We used two sampling strategies to test the predictions of the turgor limitation hypothesis.

1- Seedlings to maximum tree size within a species: We analyzed the distribution of leaf epidermal cell dimensions in different individuals of the same species, from seedlings to the tallest individuals at our study site. We selected a species that is uniformly exposed to the sun across all size classes. On the coastal plain near the Los Tuxtlas field station, Veracruz, Mexico, Bursera simaruba (L.) Sarg. (Burseraceae, Sapindales) is used as a living fence in cattle pastures. This means that the terrain is maximally open and sunny, with even small individuals exposed to full sun. We collected the apicalmost leaves from 29 individuals ranging in height from 14.5 cm to 26.6 m.

2- Trees experiencing dieback: We also study the distribution of leaf cell lengths of trees of the same species, with different heights, where the taller individuals experienced dieback. Our second strategy involved sampling a population of Eucalyptus camaldulensis Dehnh. (Myrtaceae, Myrtales) experiencing dieback in University City, of the National Autonomous University of Mexico (UNAM). As we did with Bursera simaruba, we collected the uppermost leaves of 29 individuals of Eucalyptus camaldulensis*, with heights ranging from 1.09 m to 23 m. All of the large individuals showed evidence of dieback.

We collected the uppermost leaves from each tree across all three sampling strategies. For trees under 8 meters tall, we used clippers or pruner poles, while for taller trees, we employed a DeLeaves canopy sampling system (Outreach Robotics, Quebec, Canada) mounted on a Matrice 300 RTK drone (DJI, Shenzhen, China). Tree heights were measured with a tape measure, and for the tallest trees, we utilized a Mini 4 Pro drone (DJI, Shenzhen, China) equipped with an onboard barometric altimeter.

Anatomical methods and measurement of quantitative variables: 

We preserved the leaves in 70% ethanol and then created stomatal impressions using the nail polish method. These impressions were mounted on slides, observed under a compound microscope, and photographed with a digital camera. From the stomatal impressions, we measured the lengths of guard subsidiary and pavement cells, 100 cells each, using ImageJ 1.53k, with images calibrated using a Nikon stage micrometer (MBM13100, Nikon Corp., Tokyo, Japan) to convert cell length measurements from pixels to micrometers.