We collected data in Sector Santa Rosa of the Área de Conservación Guanacaste, northwestern Costa Rica. The site is a tropical dry forest with two distinct seasons: a dry season from December through May and a wet season from June through November (Campos, 2018; Janzen and Hallwachs, 2020; Melin et al., 2020). Sector Santa Rosa forests are primarily secondary, stemming from restoration and reforestation efforts that began in the 1970s (Janzen and Hallwachs, 2020), and forest composition and structure varies between early and late successional stages. Canopy height ranges from 6 to 15 m depending on successional stage; the canopy is typically taller in areas of later succession (Kalacska et al., 2004; Powers et al., 2009).

We used cup anemometers (WL-11; Scarlet Tech, Taipei, Taiwan) to collect air speed data from May to June 2021. The instruments have a sensitivity range of 0.6–50 m/s, a resolution of 0.1 m/s, and an accuracy of ±2%, per manufacturer specifications. We built and erected two scaffold towers: one in a late successional forest (10.838617, −85.614283; 1,086 m a.s.l.), and the other in an early successional forest (10.839383, −85.616383; 906 m a.s.l.). To each tower, we affixed five anemometers at heights of 0.5, 3.5, 5.5, 7.5, and 10 m (Figure 1). The devices were set to data-logging mode, recording average and maximum air speed in 10-s intervals.

We obtained the mean, median, and standard deviation of air speed for: (1) each sampling height and (2) time of day. We calculated air turbulence as standard deviation of air speed/mean air speed (McCay, 2003). The sensitivity threshold (0.6 m/s) of our anemometer risks a systematic bias against low air speeds, especially in the understory. Accordingly, we imputed values below this limit by using survival analyses, a method developed for the health sciences but adopted for analyzing environmental data with detection limits (Helsel, 2004). Using the survival (Therneau, 2021) and NADA (Lee, 2020) packages in R (v. 4.1.0., R Core Team, 2021), we constructed Kaplan–Meier estimates for data recorded at each site independently for each height and time of day, considering air speeds ≤0.6 m/s to be left-censored.

To detect significant differences in air speeds and turbulence as a function of vertical height and time of day, we conducted Cox proportional hazard modeling via the survival package in R. As these hazard models are designed for right-censored data, we transformed our data by using the “flipping” method of Helsel (2004), which subtracts all observed values from a constant to convert left-censored data to right-censored. Since the same column of wind is recorded simultaneously at all 5 observed heights for a given site, there is a high degree of collinearity between the height and time of day predictor variables, therefore we modeled each separately and present two models. To address our first aim, we modeled air speed as a function of vertical position, based on height from the ground (“Height”) in two different habitat types. To address our second aim, we modeled air speed as a function of diel periodicity (“Time of Day”) in two different habitat types. For both models, the outcome variable was the Kaplan–Meier estimate of average wind speed, and an interaction term of study site and height, or study site and time of day, was included as a predictor variable. For the Height model, the continuous variable distance from the ground (in meters) was used, while for the Time of Day model, because of the circular nature of temporal data, we employed a sinusoidal model with separate cos and sin terms (Simmons, 1990; Cazelles et al., 2008). Code for all analyses can be found at https://github.com/allegradepasquale/wind_speed_project.git.

Air speeds increased as a positive function of vertical height, and mean air speeds differed between the study sites, with greater mean speeds recorded in the early successional forest (Figure 2 and Table 1). The interaction term of study site and height was also significant (Table 1). Variation in air turbulence followed a similar pattern: mean air turbulence scores varied as a positive function of vertical height and were greater in the early successional forest (Figure 2). Site differences in our measure of air turbulence were pronounced near ground level (0.5 m), but values tended to converge with increasing vertical height (Figure 2).

Air speed varied significantly with time of day: winds were stronger from late morning to early afternoon and were lowest overnight (Figures 3, 4). As with our analyses of height, the main effect of study site, as well as the interaction between study site and time of day, were also significant (Figure 4 and Table 1). Turbulence scores were highest at midday and lowest overnight.

We detected a steep increase in air speeds as a function of vertical height, a result that replicates findings from other forest ecosystems (Baynton, 1969; Oliver and Mayhead, 1974; Aoki et al., 1978; Kruijt et al., 2000; McCay, 2003). The prevailing explanation for this pattern is that air movement is impeded near the ground by understory vegetation, tree trunks, and surface topography; however, the implications of such a gradient for olfactory ecology are underexplored. Fruits, for example, emit odors (generally lightweight, ephemeral volatile and semivolatile organic compounds) that are easily dispersed by air movement (Rodríguez et al., 2013; Nevo et al., 2018, 2020), although excessive air speeds can over-disperse odor compounds and disrupt anemotaxis (Svensson et al., 2014). Still, it follows that the vertical position of fruit will determine the probability and efficiency of long-distance odor detection and foraging by seed-dispersing mutualists (Santana et al., 2021). Anemotaxis toward food resources is well-studied among invertebrates (Zjacic and Scholz, 2022) and increasingly so among mammals. For example, experiments with bats have elicited klinotaxis in response to fruit odors (Thies et al., 1998; Korine and Kalko, 2005; Leiser-Miller et al., 2020; Brokaw et al., 2021; Brokaw and Smotherman, 2021), and a field experiment by Fleming et al. (1977) found that bats will deviate from their flyways by as much as 50 m to acquire fruits mounted to 1.5-m poles set up moments before sunset. Further, experiments with coatis and ring-tailed lemurs have shown that they can detect fruit from distances up to 20 m (Hirsch, 2010; Cunningham et al., 2021). These findings suggest that aero-sensory ecology via stimulus response can complement and extend the critical importance of spatial memory to the localization of foods (Janson, 1998; Janmaat et al., 2014; Dahmani et al., 2018). Looking forward, it stands to reason that our understanding of cognitive ecology will only be strengthened as we incorporate and integrate the systematic study of aeroscapes and other sensory landscapes into research frameworks.

Many animals use olfactory signals to communicate with one another for reproduction, dominance, and territory defense, as well as to discriminate conspecifics from heterospecifics. In mammals, these occur mainly through the deposition of scent marks (Johnson, 1973; Irwin et al., 2004; Kollikowski et al., 2019). Our results suggest that olfactory signals may remain more localized when deposited nearer to the ground than higher in the canopy. It is noteworthy, then, that olfactory communication appears to be particularly prevalent in terrestrial mammals, which have olfactory receptor gene repertoires that have undergone three times as much gene duplication than those of volant, arboreal, and aquatic mammals (Hughes et al., 2018). At the same time, low air speeds will limit long-distance dispersal of odors. It is possible that species and habitat-specific optima for olfactory signal height exist and may vary across taxa. For example, Ethiopian wolves, gray wolves, and pine martens use raised-leg urination, a behavior thought to reinforce territorial boundaries by dispersal of urinary odor plumes. Urine deposition above, rather than on, the ground may improve scent dispersal (Peters and Mech, 1975; Macdonald, 1980; Pulliainen, 1981; Alberts, 1992; Sillero-Zubiri and Macdonald, 1998). Turning to another mammalian radiation, many primate species that have evolved glands dedicated for scent deposition, including ring-tailed lemurs, mandrills, drills, and sifakas, are largely terrestrial (Delbarco-Trillo et al., 2011; Drea, 2015; Vaglio et al., 2016). Systematic study of olfactory signaling and receiving behaviors, along with investigation of co-occurring anatomical and genetic variation, in taxa occupying different vertical niches will allow this hypothesis to be tested. In general, the intersection of sensory and aeroecology holds untapped potential for better understanding the wide variation in animal sensory traits.

We detected pronounced variation across the 24-h day, with peak and nadir air movements at midday and overnight, respectively. This pattern is almost certainly due to ambient temperature flux because warmer air moves faster and is less stable (Pleijel et al., 1996; McCay, 2003; Monteith and Unsworth, 2013; but c.f. Baynton, 1968, 1969; Rapp and Silman, 2012 for other patterns). Following similar arguments to those concerning vertical variation, nocturnal aeroscapes may be preferential for animals that detect and localize scents. Many nocturnal animals rely more on olfaction than vision, but the reasons are usually couched in the language of constraint—there is scant light at night, so vision is limited (Barton et al., 1995; Balkenius et al., 2006; Borges, 2018; Niimura et al., 2018). Underappreciated, however, is the idea that olfaction is more effective at night due to the relative stillness of air, especially in the understory, where air speeds and turbulence tend to be lowest (Murlis, 1997; Muller-Schwarze, 2006). Lack of wind produces a high signal-to-noise ratio for a given odor plume, resulting in a stronger olfactory signal. This nocturnal environment may thus select for olfactory-driven ecologies and social interactions, perhaps even in the absence of selection driven by the loss of visual ability (Drea et al., 2019).

Vertical and diel variation interact to create diverse aeroscapes, ranging from the windiest and most turbulent environment of the upper canopy at midday, to the comparative stillness of the understory at night. Such interactions create opportunities for convergence. We found that aeroscapes in the daytime understory are comparable to those in the canopy at night, which raises the possibility of convergent olfactory signals and sensitivities in the animals that occupy these distinct niches (Barton, 2006; Valenta et al., 2013; Brokaw and Smotherman, 2020; Nevo and Ayasse, 2020). Interestingly, the olfactory ecology of nocturnal terrestrial frugivores may depend in part on wind-mediated fruit falls during daytime, suggesting vertical day-night integration of aeroecologies (Augspurger and Franson, 1987; Borah and Beckman, 2021). The upshot is that daytime canopy conditions are suboptimal for anemotaxis in the service of frugivory and seed dispersal, which may explain why so-called “bird-fruits” in the upper canopy emit little scent (Gautier-Hion et al., 1985; Howe, 1986; Lomáscolo et al., 2010; Valenta et al., 2018; Valenta and Nevo, 2020).

Some angiosperm plants, including Ficus and Nicotiana, exhibit diurnal rhythms in fruit and flower chemistry (Raguso et al., 2003; Borges et al., 2011; Burdon et al., 2015; Ripperger et al., 2019; Balducci et al., 2020) timed to maximize their availability and attractiveness for pollinators and seed dispersers. These diurnal rhythms may have evolved in response to diurnal patterns in air movement, such that compounds produced during the day may be heavier and more robust against turbulent conditions than those produced at night, which may be lighter and more easily propagated under calmer conditions (Alberts, 1992; Muller-Schwarze, 2006). Flowers, which are generally more delicate and ephemeral than fruits, are often produced early in the morning, which may reflect a compromise between protection from wind-damage and availability to vision-mediated pollinators such as bees and birds (Herrera, 1990; Bloch et al., 2017). Setting aside the aeroecology of frugivory and pollination, aeroscapes are also essential to another aspect of plant reproductive biology: wind dispersal (Kennedy, 1978; Friedman and Barrett, 2008). Flowering and the eventual abscission of wind-dispersed seeds is greatest at midday, when air movement is highest, which suggests some level of aerosensation (Bohrer et al., 2008; Wright et al., 2008; Caplat et al., 2012). Such hypotheses invite future testing.

We recorded higher air speeds and greater turbulence in the early successional habitat compared to the later one, possibly reflecting the lower levels of standing biomass. This pattern is expected for habitats with greater levels of anthropogenic disturbance that have caused vegetative loss (Raynor, 1971; Muller-Schwarze, 2006; Klein et al., 2021). Such results have potentially important implications. For example, increased air movement may affect the relative colonization of wind-dispersed versus animal-dispersed plants in disturbed areas (Cadenasso and Pickett, 2001; Cubiña and Aide, 2006; Nathan et al., 2008), altering community dynamics, habitat suitability to frugivores, and reforestation efforts (Janzen, 1988; Vieira and Scariot, 2006; de la Peña-Domene et al., 2018; Camargo et al., 2020). Greater air movement and turbulence are also expected to negatively affect animals that rely on odor plumes (Shukla et al., 1990; Zhang et al., 1996; Gandu et al., 2004). Consequences include impediments to animal foraging and communication, as discussed above, as well as effects on predator-prey networks.

Sensory detection of predators often involves scent, and many species are particularly attuned to the odors of their relevant predators (Weldon, 1990; Kats and Dill, 1998; Apfelbach et al., 2005). Rats are particularly sensitive to 2,4,5-trimethylthiazoline, a component of red fox anal gland secretions (Laska et al., 2005) and mice, rats, and stoats, for example, have been shown to avoid carnivore and apex predator odors (Ferrero et al., 2011; Garvey et al., 2016). Indeed, higher wind speeds have been found to impede predator detection by mule deer (Bowyer et al., 2001) and other mammal species (Cherry and Barton, 2017). The sensory impact of air movement can be further compounded by spillover to the other senses: wind creates acoustic and visual noise, which may reduce detection of stimuli by other senses, further impeding predator detection (Hayes and Huntly, 2005; Carr and Lima, 2010; Francis et al., 2012). Overall, future work could usefully address the diverse ways that anthropogenically modified aeroscapes affect the aero-sensory ecology and habitat use of resident flora and fauna (Damschen et al., 2008). As these influences could ultimately affect species distributions (Bowyer and Kie, 2009; Breitbach et al., 2012), we urge the incorporation of aeroscapes into existing conservation and evolutionary frameworks (e.g., assessment of “edge effects” and “landscapes of fear”) and general inclusion into future studies of anthropogenic disturbance and deforestation/reforestation dynamics (Laundre et al., 2010).

Our results offer insight into the variability of air movement within a heterogeneous landscape, although some caution must be noted. First, due to equipment limitations, we were unable to sample early and late successional forest sites simultaneously, and instead sampled them sequentially. This serial approach raises the possibility that temporal differences in climate, not habitat heterogeneity, were driving the difference that we observed between sampling locations. Further, due to the logistical challenges of erecting scaffolds in a protected habitat, we were unable to sample replicates of early and late successional forest conditions. The extent to which our two study locations exemplify such habitat conditions is therefore uncertain, and could usefully be explored in greater detail in future studies sampling multiple locations per forest type. Lastly, the detection limits of our anemometers prohibited direct measurement of the slowest air speeds. While prohibitively expensive for our study design, solid-state anemometers have the advantage of greater measurement range. Future studies may also benefit from incorporating variation in wind direction, as this also has important implications for olfactory-based orientation (Kennedy, 1978; Togunov et al., 2017; Jinn et al., 2020).

Our study contributes to the emerging field of aeroecology by quantifying variation in the aeroscape of a lowland tropical dry forest and drawing attention to the implications for sensory signal propagation, plant reproduction, and mammalian sensory evolution. This topic is timely as shifting air movement patterns due to anthropogenic climate change may disrupt aeroecological interactions, with the potential far-reaching effects on animal behavior, population dynamics, and species distributions (Usbeck et al., 2010; McInnes et al., 2011; Young et al., 2011; Lewis et al., 2015). Understanding the impacts of aeroscapes on sensory ecology and evolution will be key for predicting how animals will respond to changing environments.