Mixed-vertebrate pollination traits and pollinators of Haageocereus acranthus (Cactaceae) in a Lomas desert ecosystem of coastal Peru

Bat pollination, floral traits, hummingbird pollination, intraspecific variation, morphology, 20 phenology. 21 22 Key message: In a Lomas desert ecosystem of coastal Peru, the columnar cactus Haageocereus acranthus 23 shows traits and interactions consistent with a mixed -vertebrate pollination system (hummingbirds and 24 bats), with no detectable differences between flower color morphotypes in either traits or interactions. 25 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint

26 (1) Cacti are key components of arid ecosystems and, being mostly self -incompatible, rely on animal 27 pollination. Although bee pollination is ancestral, systems supported by birds, moths, bats, and mixed 28 strategies have evolved. Despite the ecological uniquenes s and extreme seasonality of the fog -dependent 29 Lomas of coastal Peru, cacti pollination remains unstudied. This work examines Haageocereus acranthus, 30 a characteristic columnar cactus of this ecosystem, in a population where individuals produce either white 31 or pink-red flowers. It was hypothesized that white flowers would be associated with bat pollination and 32 pink-red flowers with hummingbird pollination, reflected in differences in floral morphology, phenology 33 and pollinator visitation. 34 (2) Year-round monitoring of the population was conducted to characterize flowering phenology. Floral 35 morphology, daily anthesis patterns, and nectar production were quantified and compared between color 36 morphs. Floral visitor frequency and behavior were recor ded using camera traps to test for pollinator 37 preference. 38 (3) Floral phenology, morphology, and nectar characteristics were broadly consistent with vertebrate 39 pollination, showing traits associated primarily with bat pollination but also compatible with hummingbird 40 pollination. These floral traits did not differ between color morphs. Hummingbirds were the most frequent 41 visitors, followed by bats; yet, neither group showed preference for a specific flower color. 42 (4) Findings support a mixed pollination system involving both hummingbirds and bats in an ecosystem 43 where the availability of pollinators can shift over geographic or temporal scales . This pollinator 44 unpredictability may ensure consistent reproductive success and reduce vulnerability to the absence or 45 decline of specific pollinator groups. 46 47 48 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint

49 Pollination is one of the most important mutualistic plant-animal interactions and has played a central role 50 in angiosperm diversification (Vamosi & Vamosi 2010; Ballesteros-Mejia et al. 2016). These interactions 51 reflect the evolutionary innovations of flowers in response to different ecological conditions (Armbruster 52 2017; Opedal 2019). In this context, species sharing pollinators have been under similar selective pressures, 53 resulting in a convergence of floral traits known as pollination syndromes (Faegri & van der Pijl 1979; 54 Fenster et al. 2004; Dellinger 2020). 55 Pollination syndromes have frequently been considered a principle governing plant -pollinator 56 interactions and used to infer functional pollinator groups based on flower traits (Proctor et al. 1996; Fenster 57 et al. 2004). However, as the understanding of pollination has expanded, increasing evidence supports that 58 these patterns are more complex than the traditional syndromes hypothesis suggests (Ollerton et al. 2009; 59 Rosas-Guerrero et al. 2014; Dellinger 2020). Pollinator assemblages can be highly variable across the 60 distribution of a particular plant species (Thompson 2002), and interactions often show geographical or 61 temporal changes (Olesen & Jordano 2002; Burkle & Alarcon 2011). Also, intraspecific variation in floral 62 traits may contribute to deviations from consis tent classic syndromes, as different morphotypes within a 63 species distribution range can attract distinct pollinators groups (e.g., Schlumpberger et al. 2009; Cardona 64 et al. 2020; Wenzell et al. 2025). These geographic patterns in interactions, along with intraspecific floral 65 variation, can drive local adaptation, divergence in pollination strategies, and potentially lead to speciation 66 (Kay & Sargent 2009). 67 Cacti are distributed across the Americas and constitute crucial components of arid and semi -arid 68 ecosystems (Fleming et al. 2001; Fleming & Valiente-Banuet 2002). In addition, these are obligate out -69 crossers that depend on animals for pollination (Mandujano et al. 2009) and show adaptations mostly 70 consistent with classic pollination syndromes (Grant & Grant 1979; Rowley 1980; Schlumpberger 2011). 71 While bee pollination is ancestral in cacti, specialized systems for bird, moth, and bat pollination have 72 evolved; including mixed pollination strategies (Schlumpberger 2011; Lendel 2013 ) (e.g., Sahley 1996; 73 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Bustamante et al. 2010; Walter 2010). This diversity of strategies makes cacti an excellent group to explore 74 and test hypotheses on pollination syndromes. 75 Peru is a cacti species -rich country with ~250 spp. distributed mainly in coastal and Andean 76 ecosystems (Ulloa et al. 2004; Arakaki et al. 2006). However, although Peruvian cacti have received 77 considerable attention at taxonomic, distributional and genetic studies (e.g., Ostolaza 1996, 2014; Arakaki 78 et al. 2006, 2007, 2021; Calderón et al. 2007), scarce effort has been made to understand the ir ecological 79 interactions (e.g., Sahley 1996; Novoa et al. 2005, 2022; Ceroni et al., 2007). 80 Haageocereus acranthus (Tribe Trichocereeae) is a representative and abundant member of the 81 genus, occurring predominantly in arid river valleys of Peru (Calderón et al. 2007). Flowers of H. acranthus 82 are relatively large and robust, funnel - to tubular- shaped, single-opening, nocturnal, white colored and 83 nectar-abundant. These traits have been putatively associated with bat and sphingid moth pollination by 84 Calderón et al. (2007), based on classic syndromes proposal by Faegri & van der Pijl (1979). However, this 85 species exhibits high variability in floral traits, including flower size, shape, and pigmentation ranging from 86 white to pink-red. This variation, along with extended anthesis times (crepuscular and matutinal), suggests 87 potential pollination by other groups, such as hummingbirds. While Rodophis vesper (Oasis hummingbird) 88 and Platalina genovensium (Peruvian long-tongued bat) have been reported visiting Haageocereus flowers 89 (Grillo & Arana 2016; Maguiña & Amanzo 2016), no further studies have an alyzed these interactions in 90 detail. Also, c omparable traits occur in Weberbauerocereus weberbaueri , a closely related species 91 pollinated by both bats and hummingbirds on the southern Peruvian coast (Sahley 1996). 92 This study examined the floral morphology, phenology, and pollinator assemblage of 93 Haageocereus acranthus in a Lomas ecosystem of the Peruvian central desert coast, where the species 94 exhibits two distinct flower colors: white and pink-red, each expressed by different individuals. We 95 hypothesized that H. acranthus flowers would generally display traits associated with ve rtebrate 96 pollination, particularly by bats and hummingbirds, and that these visitors would act as effective pollinators. 97 Based on classic pollination syndromes, we further predicted intraspecific differences linked to flower 98 color: white flowers would be more closely associated with bat pollination, whereas pink-red flowers would 99 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint show a stronger tendency toward hummingbird pollination. We anticipated that these flower types would 100 occupy distinct regions of floral morphological space and exhibit different phenological patterns, with pink-101 red flowers showing a broader anthesis and nectar secretion period that extend s into diurnal hours. 102 Consequently, we also expected differences in visitation frequency between bats and hummingbirds across 103 flower types. 104

105 Study area and species 106 This study was conducted in Cardal (Pachacamac, Lima, Peru) (12°11'S, 76°50'W; 215 m a.s.l.), a location 107 with a previously documented H. acranthus population (Ostolaza 1996; Calderón et al. 2007) (Fig. 1A). 108 This area is part of the Lomas, a unique ecosystem dependent on winter fogs, which is mainly restricted to 109 coastal Peru and Chile, and characterized by two contrasting seasons. The wet season (May –Oct), with 110 lower temperatures (13℃–16℃) and high relative humidity (95%–100%), promotes the growth of seasonal 111 herbs. In contrast, during the dry season (Nov –Apr), with relative humidity ranging from 80% –90% and 112 temperatures between 20℃ –25℃, most herbs die off, leaving only perennial xerophytes ( Dillon et al. 113 2011). Fieldwork was conducted from January 2022 to March 2023. 114 H. acranthus is a columnar cactus up to 1.5 m tall, with multiple vertical stems branching at the 115 base and flowers often growing from areoles at the branch apex (Calderón et al. 2007) (Fig. 1B). Aside 116 from observations of flowering from November to January, no detailed phenological information is 117 available (Calderón et al. 2007; Maguiña & Amanzo 2016 ). Two flower types are present in this area: 118 typical white flowers of H. acra nthus subsp. acranthus and pink-red flowers of H. acranthus var. 119 olowinskianus f. rubriflorior (Ostolaza 1996), currently a synonym of H. acranthus subsp. acranthus 120 (Calderón et al. 2007; POWO 2024). Flower types are hereafter referred to as white and pink-red floral 121 morphs. 122 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Although H. acranthus dominates the site, other cacti co-occur at much lower densities, including 123 H. pseudomelanostele (Werderm. & Backeb.) Backeb. and Loxanthocereus acanthurus (Vaupel) Backeb. 124 The former is morphologically and phenologically similar to H. acranthus (Ostolaza 1996; Calderón et al. 125 2007), whereas the latter exhibits a hummingbird-pollination syndrome, flowering at a different time of the 126 year (B. Garcia-Simpson, pers. obs.). 127 Annual phenology 128 We monthly monitored 30 tagged individuals throughout one year (Jan –Dec 2022). For every individual, 129 we recorded the number of floral buds, open flowers and fruits (unripe and ripe). Unripe and ripe fruits 130 were differentiated by coloration and firmness (firm and green for unripe, red and soft for ripe). Due to field 131 limitations, monitoring was conducted exclusively on individuals with white flowers. However, qualitative 132 observations from recurrent prior and subsequent visits to the study area (years 2021 to 2025; B. Garcia-133 Simpson, pers. obs.) suggest that the phenological pattern of the white morphotype is representative of the 134 overall population. 135 Floral traits – morphology 136 We randomly collected 15 fully open flowers of each floral morph from different individuals. Longitudinal 137 cut sections were photographed in the field with an iPhone XR camera (12MP, f/1.8 aperture, 26mm focal 138 length) (Apple, California, USA) against a black background with a reference scale. Total length (TL), 139 perianth width (PW), tube length (TuL), tube width (TuW), stigma exsertion (StiE), stamen exsertion 140 (StaE), ovary length (OL), ovary width (OW), nectar chamber length (NL), and nectar chamber width (NW) 141 (Fig. 2A–D; terminology modified from Nassar et al. 1997) were measured from photos using ImageJ 142 v.1.54f (Schneider et al. 2012). 143 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Floral traits – daily phenology 144 To characterize flower aperture and nectar production, we measured the perianth width of 23 flowers ( 11 145 white, 12 pink-red) from different individuals at eight stages of anthesis (15:00, 17:00, 19:00, 23:00, 03:00, 146 05:00, 07:00, 09:00) using a 0.05 mm resolution mechanical caliper (Uyustools, Hangzhou, China). Nectar 147 volume, sugar concentration, and energy content were assessed at five anthesis stages (15:00, 19:00, 23:00, 148 03:00, 06:00) using bagged flowers from different individuals. Nectar volume was measured in 34 flowers 149 (18 white, 16 pink-red), while sugar content and energy supply were evaluated from a subset of 17 flowers 150 (10 white, 7 pink-red). Nectar volume was quantified, following Ibarra-Cerdeña et al. (2005), by extracting 151 all possibl e nectar using a capillary tube inserted into the nectar chamber. Total nectar volume was 152 calculated by multiplying the column length by the specified cross-sectional area of the capillary tube. A 153 fraction of the nectar was used to determine sugar concentr ation with a handheld refractometer (BRIX30) 154 with automatic temperature compensation; readings were expressed as sucrose percentage following Dafni 155 (1992) and energy supply was calculated using the following formula: 156 𝐽 = 16.8[( 𝑆 100) ∗ 𝑉𝐷] 157 where J is energy in joules, S is the sugar percentage, V is nectar volume in microliters, and D is the density 158 of sucrose at the observed concentration (Dafni 1992; Ibarra-Cerdeña et al. 2005). 159 Frequency and behavior of floral visitors 160 We monitored floral visitors using two Bushnell 24MP Core Low Glow Trail Cameras (Bushnell 161 Corporation, Kansas, USA) placed 1 –2.5 meters from flowers to capture interactions with vertebrate and 162 large invertebrate visitors. A total of 22 ( 11 white, 11 pink-red) flowers from different individuals were 163 observed over 11 non -consecutive days during the flowering season (Sep 2022 –Mar 2023), totaling 378 164 observations hours (~17 hours per flower). Monitoring started in the early afternoon and continued until 165 mid-morning the following day. Based on Ibarra -Cerdeña et al. (2005), we recorded visitor species and 166 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint number of legitimate visits (i.e., involving contact with reproductive structures) for each flower. Additional 167 visitors were documented through direct observation throughout anthesis. 168 Preliminary pollinator exclusion experiments 169 To evaluate how the absence of different pollinator groups influenced fruit initiation, we implemented five 170 treatments: (i) control, with flowers left fully exposed; (ii) bat exclusion, covering flowers at night with a 171 cylindrical metallic mesh; (iii) hummingbird exclusion, applying the same mesh only during the daytime; 172 (iv) nocturnal exclusion, fully bagging flowers at night to prevent all nocturnal visitors; and (v) diurnal 173 exclusion, fully bagging flowers during the daytime to block diurnal visitors. We monitored each flower 174 over the following days to record whether it abscised or initiated ovary expansion. Due to fieldwork 175 constraints, sample sizes were limited to 12 flowers per morph in the control treatment and six flowers per 176 morph in each exclusion treatment. Therefore, results are presented descriptively as preliminary evidence, 177 without formal statistical analyses. 178 Data analyses 179 All statistical analyses were conducted in R v.4.4.1 (R Core Team 2024) using RStudio (Posit Team 180 2024). 181 Intraspecific morphological variation. We evaluated the effect of flower morph (white vs pink-red) on 182 all morphometric variables using a MANOVA (Pérez -Barrales et al. 2007). Variables were standardized 183 (mean = 0, standard deviation = 1) for comparability. To avoid multicollinearity, we examined correlations 184 among morphometric variables: total length, tube length, and nectar chamber length were highly correlated 185 (Pearson’s r > 0.8), therefore, we excluded tube length and nectar chamber length from the test. The analysis 186 was performed with the manova function. 187 Modelling flower anthesis and nectar production. We fitted Generalized Linear Mixed-Effects 188 Models (GLMMs) to examine the effects of time after midday (hours) and flower morphotype (white vs 189 pink-red) on four response variables: (i) flower aperture, (ii) nectar volume, (iii) nectar sugar concentration, 190 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint and (iv) nectar energetic content. In the absence of specific hypotheses regarding differences among 191 morphotypes (e.g., response magnitude or peak times), we opted for an exploratory model selection 192 approach by generating a set of biologically plausible candidate models and selecting the best fit. 193 Model selection was conducted separately for each response variable. Candidate models were built 194 using the glmmTMB package (Brooks et al. 2017), with Flower ID included as a random effect to account 195 for individual variability. Time after midday (T) and flower morphotype (M) were included as fixed effects, 196 testing first-, second-, and third-order orthogonal polynomial terms for time to capture potential curvilinear 197 relationships. Fixed effects combination in candidate models included: T, T + M, T * M, T 2, T2 + M, T2 * 198 M, T 3, T 3 + M, T 3 * M. When a model includes a polynomial of degree n, it inherently incorporates all 199 lower-order terms (e.g., a T³ model also includes T and T²). Here, ‘+’ indicates additive effects, while ‘*’ 200 denotes additive and interaction effects. Flower aperture and sugar content were modeled using a Gaussian 201 distribution, while nectar volume and energetic content were modeled with a Tweedie distribution, chosen 202 for its suitability in handling continuous data with a high proportion of zeros. Model diagnostics; presence 203 of over/under-dispersion, outliers, and zero inflation, were assessed using the DHARMa package (Hartig 204 2022). Model selection indices were computed using compare_performance from the performance package 205 (Lüdecke et al. 2021) and model.sel from the MuMIn package (Barton 2024). Final selection was based on 206 the Akaike Information Criterion cor rected for small sample sizes (AICc), with models differing by ≥ 2 207 AICc units considered significantly better fits (Burnham & Anderson 2002). 208 Frequency and behaviour of floral visitors. To visualize the temporal activity patterns of main 209 floral visitors, we used a kernel density estimation (KDE) with the density function (bandwidth = 1). 210 Densities were scaled by the total number of visits to make activity patterns visually comparable across 211 visitor groups with differing number of visits. 212 To dete rmine if the number of visits varied among floral visitors and flower types, we built 213 Generalized Linear Mixed-effects Models (GLMMs) with a Poisson distribution (log-link). Flower ID was 214 considered as a random effect, while visitor type (bat, hummingbird or sphingid) and flower morphotype 215 were considered as fixed effects. We evaluated three models including effects of (i) visitor type (V), (ii) 216 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint visitor type and flower morphotype (V + M), and (iii) visitor type, flower morphotype, and their interaction 217 (V * M). To account for differences in evaluation time among individual flowers, all models included the 218 log-scaled evaluation time of each flower as an offset variable. Model fitting, diagnostics, and selection 219 were performed as previously described. Based on the best -supported model, post-hoc pairwise 220 comparisons of visitation frequencies among floral visitors were conducted using the emmeans package 221 (Lenth & Lenth 2018). Pairwise differences were evaluated with the pairs function, applying a Tukey 222 adjustment to account for multiple comparisons. 223

224 Annual phenology 225 H. acranthus from Cardal displayed a single flowering peak during the year. Bud and flower production 226 occurred primarily from January to April, followed by a sharp decline. From May to September, plants 227 remained mostly vegetative with minimal reproductive activity. Bud production resumed in October, 228 peaking in November, resulting in a high number of open flowers. Fruit production followed this pattern 229 but with a slight delay due to the s equential nature of these reproductive stages. Overall, reproduction 230 occurred during the dry season (Nov–Apr), while the vegetative phase in the humid season (May–Oct) (Fig. 231 3A–B; Table S1). 232 Floral traits – morphology 233 H. acranthus flowers were funnelform to tubular in shape and radially symmetric , but occasionally 234 exhibited a certain degree of zygomorphy (Fig. 2B). The flower tube exhibited a constriction above the 235 nectar chamber, which limits access to nectar. Perianth color varied from white and white/green to pink-236 red, but remained consistent within individuals (Fig. 2C–D). Overall morphometry of the two floral morphs 237 did not differ (Table S2), as supported by the MANOVA results (Pillai’s Trace = 0.221, F (8, 23) = 0.813, 238 P = 0.598). 239 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Floral traits – daily phenology 240 Unless otherwise indicated, model estimates are reported with 95% confidence intervals (mean ± 95% CI). 241 Flower aperture. The top-ranked model included time up to the cubic term (T, T², T³) and its 242 respective interactions with morphotype (AICc = 467.29, weight = 0.560). The second -best model 243 additionally included the main effect of morphotype (ΔAICc = 2.31, weight = 0.176) (Table S3). The strong 244 second-order polynomial effect confirmed the parabolic pattern in flower aperture over time, showing the 245 expected non-linear pattern of anthesis. The interaction terms, present in the final model but with a weak 246 effect, revealed that this parabolic relationship varied between morphs, highlighting slight differences in 247 how aperture responds to time. Both floral morphs reached maximum aperture around 23:00 (white: 6.51 248 ± 0.3 cm, pink-red: 5.9 ± 0.3 cm) and differences between morphs were observed in the early stages of 249 anthesis, where the pink-red morph showed higher estimates than the white morph, indicating a slightly 250 earlier aperture initiation (white: 14:34, pink-red: 12:81) (Fig. 4A, Table 1A). Visually, styles and stigmas 251 remained turgid throughout anthesis. 252 Nectar volume. The best model for nectar volume included time up to the third-order polynomial 253 term but excluded interactions with morphotype and its main effect (AICc = 1025.65, weight = 0.920). The 254 second-best model included interactions with morphotype (ΔAICc = 5.46, weight = 0.060). The main effect 255 of morphotype was present in models of lower rank (Table S 4). The second- and third-order polynomial 256 terms together indicated a parabolic pattern in nectar volume over time, with an initial rise in late afternoon 257 to a peak up to 93.5 ± 13.5 µL at 21:00 followed by a less steep decline, where lower quantities of nectar 258 were available until the following morning (Fig. 4B, Table 1B). 259 Sugar content. The three best -ranked models showed ΔAICc < 2 among them ; we selected the 260 simpler model, which included the linear effect of time without the interaction with morphotype nor its 261 main effect (AICc = 164.00, weight = 0.265). This model showed equivalent support compared to more 262 complex models that included the second and third-order effect of time (AICc = 163.28, weight = 0.379), 263 or the main effect of morphotype and its interactions (AICc = 165.08, weight = 0.154) (Table S 5). The 264 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint model estimates an initial content of sugars of 23.5 ± 0.5 % at 18:30, showing a linear decline over time, 265 reaching lower values of 15.4 ± 0.8 % at 07:40 (Fig. 4C, Table 1C). 266 Energy content. The best model for nectar energ y content included time up to the third -order 267 polynomial term but excluded morphotype effects (AICc = 540.06, weight = 0.738). The second -best 268 model, which added interactions with morphotype, had less support (ΔAICc = 3.05, weight = 0.160). The 269 main effect of morphotype was included in models of lower rank (Table S6). Energetic supply followed a 270 similar pattern to nectar secretion, with an initial rise in late afternoon to a peak up to 529.8 ± 77.8 J at 271 20:30 followed by a less steep decline lasting until the following morning (Fig. 4D, Table 1D). 272 Frequency and behavior of floral visitors 273 We recorded 11 invertebrate and two vertebrate species of floral visitors; two nocturnal, five diurnal, and 274 six active during both periods (Table 2). Nine did not contact the reproductive structures, making them 275 unlikely pollinators due to their incompatibility with floral morphology. Two species occasionally made 276 contact with the reproductive structures, while the remaining two consistently made eff ective pollinator 277 visits, regularly contacting both reproductive structures while accessing floral resources (Table 2). 278 Among the small flower visitors, two ant species (Formicidae) were observed visiting flowers at 279 night, primarily for nectar and often in large numbers (> 20 individuals). These ants continued harvesting 280 nectar until the flowers senesced (Fig. S1A–C). One sap beetle species (Carpophilus sp., Nitidulidae) was 281 found inside flowers, likely feeding on pollen or plant tissue, and was more abundant in older flowers (Fig. 282 S1D). Three spider species (Anyphaenidae, Thomisidae , and Salticidae) were recorded on the flowers, 283 occupying the external zones of the floral tubes and perianth, acting as predators (Fig. S1E–F; Fig. S2A–284 B). Two Diptera species were documented; one was abundant and found inside the floral tubes (Phoridae) 285 (Fig. S2C), and a syrphid was observed once feeding on pollen. The common bee ( Apis mellifera) was 286 frequently observed collecting pollen from anthers but rarely contacting the stigma, except when gathering 287 in larger groups (> 5 individuals) (Fig. S2D), while the metallic-green bee Caenohalictus sp. (Halictidae) 288 was also recorded (Fig. S1E–F). 289 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Regarding large flower visitors, the hummingbird Rhodopis vesper (Trochilidae) was the most 290 frequent (Fig. 5 A–B). Their activity started in the late afternoon (17:00) until sunset (18:30 –19:00) and 291 resumed in the early morning (05:00) until the end of anthesis (9:00–11:00) (percentage of total visits: 292 white: 71.4%, pink-red: 81.4%). They hovered or, in occasions landed on flowers to access nectar, always 293 contacting stamens and stigma. The Peruvian Long-tongued Bat, Platalina genovensium (Phyllostomidae), 294 was the second most frequent visitor (Fig. 5C–D), active at night (20:00–03:00), peaking at 22:00 (white: 295 20.6%, pink-red: 16.3%). Unlike R. vesper, bats made single contacts to flowers while hovering but never 296 landed on them. Visits by h awkmoths (Manduca sp., Sphingidae, Lepidoptera) were rare and recorded at 297 dusk or at night (white : 7.9%, pink-red: 2.3%). They either hovered or landed on flowers, but their body 298 not always contacted reproductive structures (illegitimate visits = 2, legitimate visits = 4) (Fig. 5E–F). 299 The model including only the effect of flower visitor was selected as the best-fitting model (AICc 300 = 209.6, weight = 0.717). The additive (ΔAICc = 2.32, weight = 0.225) and multiplicative models received 301 significantly less support ( ΔAICc = 5.05, weight = 0.057) (Table S 7). In consequence, we compared 302 visitation frequencies among visitors based on the first model (Table 1E); which showed sphingids had 303 significantly lower visitation rates compared to both P. genovensium (estimate = -1.20, SE = 0.465, P = 304 0.0262) and R. vesper (estimate = -2.59, SE = 0.423, P < 0.001). Additionally, P. genovensium exhibited 305 lower visitation rates than Rhodopis vesper (estimate = -1.39, SE = 0.250, P < 0.001). 306 Preliminary pollinator exclusion experiments 307 Control flowers showed the highest fruit initiation rates (white: 41.7%; pink -red: 33.3%). Bat exclusion 308 reduced fruit initiation in the white morph but not in the pink-red morph (white: 16.7%; pink-red: 33.3%). 309 Hummingbird exclusion lowered fruit initiation in both morphs (white: 16.7%; pink-red: 16.7%). Nocturnal 310 exclusion produced moderate initiation in the white morph but low initiation in the pink-red morph (white: 311 33.3%; pink -red: 16.7%), and diurnal exclusion resulted in the same pattern (white: 33.3%; pink -red: 312 16.7%). In general, fruit initiation was low across all treatments. When morphs were pooled, most 313 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint treatments showed a fruit initiation rate of 25%, except for the hummingbird -exclusion treatment, which 314 showed 17%. The control flowers exhibited the highest fruit initiation rate at 37.5%. 315

316 Annual phenology 317 The Lomas formations are marked by strong seasonality in floral resources availability, with most plant 318 species reproducing during the wet season (Tovar et al. , 2018). In contrast, our data show s that 319 Haageocereus acranthus flowers in the dry season, with a single reproductive peak. Similar dry -season 320 flowering has been reported at Lachay National Reserve by Maguiña & Amanzo (2016), and Calderón et 321 al. (2007) describe flowering in January and fruiting in February. This recurring pattern across different 322 Lomas populations of H. acranthus suggests that rising temperatures (Servicio Nacional de Meteorología 323 e Hidrología del Perú - SENAMHI; Figure S1), and reduced humidity may act as cues for bud development, 324 although this requires further testing. Flowering during resource-limited periods positions H. acranthus as 325 a potential key species in Lomas ecosystems, providing an important resource for associated fauna. 326 Floral morphology, phenology, and expected pollination syndrome 327 The floral dimensions of H. acranthus are consistent with those of other vertebrate -pollinated cacti. Its 328 large, sturdy flowers can withstand landings and manipulations by relatively large visitors, while the broad 329 frontal area (~6.3 cm diameter) likely enhances detection by bats through echo location, as shown for 330 Pachycereus (González-Terrazas et al., 2016). This surface also provides ample contact for pollen transfer. 331 Each flower bears 200 –400 stamens (Calderón et al., 2007), indicating high pollen output and potential 332 seed set. Comparable b at-pollinated species produce 10⁵ –2×10⁵ pollen grains per flower (Nassar et al., 333 1997), with Pilosocereus reaching ~500 stamens, ~2000 ovules, and ~1100 seeds per fruit (Martins et al., 334 2020). Together, these traits suggest that H. acranthus depends on pol linators with high pollen -carrying 335 capacities to ensure reproductive success within its short floral lifespan (< 24 h). 336 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint The floral tube is determinant in cacti plant-pollinator interactions (Schlumpberger 2011). In H. 337 acranthus, an average tube length of 4.8 cm restricts nectar access to visitors with sufficiently long feeding 338 structures, establishing a threshold for effective foraging. Some species with shorter feeding structures can 339 still exploit the resource by inserting their head or rostrum into the funnel-shaped tube, which allows deeper 340 access than in narrower flowers (e.g., R. vesper; Fig. 5B). Experimental studies further show that tube width 341 influences foraging behavior: wider tubes extend the feeding reach of nectar -feeding bats (Winter & von 342 Helversen 2003; Nicolay & Winter 2006), whereas narrower tubes increase hummingbird handling times 343 (Smith et al. 1996). Finally, visitors with feeding structures exceeding ~6.8 cm (tube length plus stamen 344 exsertion) may obtain nectar without contacting the rep roductive organs, thereby acting as nectar robbers 345 (e.g., sphingid moths; Fig. 5F). 346 Monitoring of flower phenology confirmed that H. acranthus flowers open only once, with peaks 347 in aperture, nectar secretion, and energy availability occurring at night (Fig. 4A –D). Total nectar 348 production, sugar content, and energetic supply were high, comparable to values reported for cacti 349 pollinated by hummingbirds and bats (e.g., Rowley 1980; Grant & Grant 1980; Schlumpberger et al. 2009, 350 Schlumpberger 2011; Albuquerque-Lima et al. 2023). Although flower opening varied among morphotypes 351 (pink-red flowers opened slightly earlier than white flowers), nectar characteristics did not differ, providing 352 insufficient evidence for phenological divergence among morphs. These ch aracteristics suggest an 353 association with pollination by nocturnal vertebrates or large invertebrates (Baker 1975; Medel et al. 2022). 354 However, anthesis extended into the late afternoon and morning, whe n small amounts of nectar were 355 available (Fig. 4B), allowing diurnal vertebrates to forage flowers and potentially contribute to pollination 356 (Baker 1961; Miyake & Yahara 1999). 357 Although pink-red and white pigmentation are traditionally linked to diurnal hummingbird and 358 nocturnal bat/sphingid pollination, respectively, our analyses did not reveal differences among white and 359 pink-red floral morphs. Besides slight differences between flower aperture patterns, floral morphology and 360 phenology were consistent across morphs. Overall, the observed floral traits strongly support the hypothesis 361 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint of primary adaptation to bat pollination, with characteristics also associated to hummingbird pollination. In 362 contrast, the relatively short floral tube indicates only limited adaptation to sphingid pollination. 363 Empirical evidence of flower visitors and pollinators 364 Visitors are here grouped into three categories based on morphology, behavior, and contact with 365 reproductive organs: (1) ants, sap beetles, flies, and arachnids, whose foraging rarely involves stigma or 366 stamen contact and thus contribute little to no pollination; (2) bees and sphingid moths, which may 367 occasionally transfer pollen but more often act as nectar or pollen thieves; and (3) vertebrates, whose traits 368 align more closely with floral morphology and are therefore expected to be the most effective pollinators. 369 Ants, sap beetles, flies and arachnids. Ants were frequently observed in large groups on flowers, 370 but their small size, unlikely stigma contact, and foraging behavior suggests negligible contributions to 371 pollen transfer (Rico -Gray 1989; Fagua & Ackerman 2011; LeVan et al. 2014). Similarly, Carpophilus 372 beetles, common in both wild and cultivated cacti, ac t as pollen thieves they feed on nectar, lay eggs in 373 buds, and their larvae consume decaying flowers (Grant & Connell 1979; Miranda -Jácome et al. 2021). 374 Dipterans frequently visited H. acranthus flowers but rarely contacted r eproductive organs due to their 375 small size, making them ineffective pollinators (Rowley 1980; McIntosh 2005; Schlumpberger et al., 2009). 376 Finally, spiders were exclusively predatory visitors, typically using flowers as hunting grounds (Su et al. 377 2020), though occasional resource use has been reported (e.g., Nelson 2023). 378 Bees and sphingid moths. Bees are frequent visitors due to their pollen -collecting behavior 379 (Westerkamp 1996; Thorp 2000) and, when morphologically compatible, can be effective cross-pollinators 380 (Armbruster et al. 1989; Solís‐Montero & Vallejo‐Marín 2017). In H. acranthus, however, the large flowers 381 and exserted stigmas limit bee -stigma contact, which could reduce their effectiveness as pollinators. 382 Although not quantified in our study, studies on tropical cacti with similar floral traits and growth habit 383 suggest that the contribution of bees and other small insects to pollination is low or null (e.g., bees, flies, 384 and butterflies on Weberbauerocereus in Sahley 1996; A. mellifera on Pilosocereus in Rivera-Marchand & 385 Ackerman 2006; Xylocopa grisescens on Pilosocereus in Rocha et al. 2019; A. mellifera on Cipocereus in 386 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Martins et al. 2020). In contrast, in temperate zones bees often act as effective pollinators, even in species 387 with floral traits associated with bats or moths, reflecting more generalized pollination strategies (Valiente-388 Banuet et al. 1996; Fleming et al. 2001). 389 The traits of H. acranthus may also suggest sphingid moth pollination (e.g., white pigmentation, 390 nocturnal anthesis, nectar rewards). However, sphingids accounted for less than 10% of visits and were 391 observed both landing on flowers or accessing nectar without contacting the stamens or stigma. This 392 behavior has been previously documented in moths, as it allows t hem to feed from the relatively short -393 tubed flowers of while avoiding close approaches that increase predation risk (Wasserthal 2001). By 394 contrast, species adapted to sphingid pollination usually possess much longer floral tubes (e.g., > 10 cm in 395 Grant & Grant 1979; > 15 cm in Silva & Sazima 1995; > 15 cm in Schlumpberger et al. 2009; ~24 cm in 396 Albuquerque-Lima et al. 2023). Thus, the role of hawkmoths as pollinators of H. acranthus results unlikely. 397 Bats and hummingbirds. Hummingbird foraging overlapped with anthesis only during 398 crepuscular and early morning hours, yet their visits accounted for a major proportion of total observations. 399 The high frequency of R. vesper visits highlights its potential role as a pollinator, ev en though nectar 400 availability at these times was often low or undetectable . This can be explained by the fact that nectar 401 measurements involved repeated extraction throughout anthesis; therefore, morning measurements 402 reflected nectar levels under continuous extraction rather than natural overnight accumulation and cannot 403 rule out nectar persistence in flowers unvisited during the night. 404 In Cactaceae, hummingbird -pollinated flowers are typically red to orange, less robust, 405 zygomorphic, and tubular (e.g., Loxanthocereus, Borzicactus, Matucana spp.) ( Grant & Grant 1979; 406 Rowley 1980; Schlumpberger 2011). Although the morphology of H. acranthus does not fully match this 407 bauplan, it exhibits secondary compatibility with R. vesper to access nectar while ensuring contact with 408 reproductive structures. Visitation rates did not differ between morphs, indicating that both floral color 409 variants are equally attractive to hummingbirds. Overall, the frequent and effective visits of R. vespe r 410 support its role as a compatible pollinator, consistent with reports of this species visiting Peruvian cacti 411 lacking typical ornithophilous traits (Sahley 1996; Novoa et al. 2022). 412 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint The synchronization of H. acranthus phenology with the nocturnal activity of P. genovensium, 413 coinciding with peak flower aperture and nectar availability is consistent with bat pollination (e.g., Valiente-414 Banuet et al. 1996; Nassar et al. 1997; Ibarra-Cerdeña et al. 2005; Rocha et al. 2019; Martins et al. 2020; 415 Albuquerque-Lima et al. 2023). Morphologically, P. genovensium is well-suited to exploit these flowers, 416 as its long feeding structures match the floral tube. Observations of individuals carrying Haageocereus 417 pollen (Maguiña & Amanzo 2016), along wit h its status as a cactus specialist (Sahley & Baraybar 1996), 418 further support its role as a pollinator. Although bats were not the most frequent visitors, they exhibited the 419 greatest morphological fit to the flowers, suggesting higher pollen transfer efficiency per visit compared to 420 hummingbirds (Muchhala & Thomson 2010) . Thus, P. genovensium may represent the most effective 421 pollinator of H. acranthus, as Leptonycteris yerbabuenae does for several columnar cacti in North America 422 below 22º latitude (e.g., Tremlett et al. 2020). 423 Preliminary pollination exclusion experiments in other Lomas (Lachay, northern Lima) suggest 424 that H. acranthus is primarily pollinated at night (Grillo & Arana 2016). In Lachay, hummingbirds visited 425 flowers frequently, but bat activity was slightly higher. Combined with our observations, this evidence 426 indicates that H. acranthus is mainly adapted to bat pollination, though the predominance of chiropterophily 427 may depend on the presence of P. genovensium. This bat species is highly dependent on cactus flowers and 428 intact arid habitats, and its distribution is restricted to well -preserved areas that are rapidly declini ng in 429 Lima (Ossa et al. 2020; Lambert 2021). Accordingly, P. genovensium has not been reported from urban 430 environments, where only generalist bats persist by feeding on exotic plants (Pellón et al. 2021). The peri-431 urban setting of our study site, surrounde d by towns and agriculture, contrasts with Lachay, a conserved 432 National Protected Area (Dourojeanni 2018), likely explaining the greater prevalence of hummingbird 433 visitation at Cardal. In addition, our preliminary exclusion experiments showed that , although fruit 434 production been generally low, all exclusion treatments tended to produce equal or lower fruit initiation 435 than open flowers, suggesting that both diurnal and nocturnal pollinators contribute to fruit set. Together, 436 these patterns support that the m ixed-vertebrate pollination system of H. acranthus may facilitate its 437 persistence even where its ideal bat pollinators decline or are absent. 438 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Our findings partially align with those of Sahley (1996) for Weberbauerocereus weberbaueri in 439 Arequipa (southern Peru). This closely related species (Tribe Trichocereeae) shares similar morphological 440 and phenological floral traits and is also primarily visited by P. genovensium and R. vesper (although in W. 441 weberbaueri the wine-colored flowers are strongly z ygomorphic, unlike in H. acranthus). Sahley (1996) 442 documented interannual shifts in pollinator dominance, with bats prevailing in one year and hummingbirds 443 in other, a pattern linked to El Niño events that alter rainfall, vegetation, and bat abundance. We propose 444 that H. acranthus may follow a similar strategy, with mixed floral traits enabling dual pollination. Such 445 flexibility is likely advantageous in the seasonal Lomas ecosystem, where extreme climatic events like El 446 Niño strongly influence resource availability (Ferreyra 1993; Cano et al. 1999). The hypothesized migratory 447 behavior of P. genovensium (Sahley 1996; Sahley & Baraybar 1996; Ossa et al. 2020) further underscores 448 the value of maintaining alternative pollinators: hummingbirds may provide reliable service during periods 449 of reduced bat activity, thereby supporting reproductive success under fluctuating conditions (Muchhala & 450 Thomson 2010). Finally, the disjunct distribution of H. acranthus along the Peruvian coast, with 451 populations separated by large distances, aligns with bat-mediated pollination, as bats are highly effective 452 long-distance pollen vectors, promoting genetic connectivity despite habitat fragmentation (Fleming et al. 453 2009). 454 Mixed-pollination strategies in a broader context 455 The most effective pollinator principle states that floral traits should evolve toward the selective optimum 456 imposed by the pollinator contributing the most to reproductive success ( Stebbins 1970), but this is not 457 always observed. When no single guild imposes strong or consistent selection, floral phenotypes matching 458 multiple guilds can be favored because their combined contributions exceed those provided by 459 specialization. Such mixed strategies often emerge when main pollinator abundances, and thus their relative 460 contributions to plant fitness, shift across space or time (Kay & Anderson, 2025). 461 In some cases, mixed pollination occurs even in species with clear floral syndromes. For example, 462 Marginatocereus marginatus in Mexico , although exhibiting traits associated with hummingbirds, 463 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint competition with co-flowering columnar cacti reduced the reliability of individual pollinator guilds, and its 464 dual nocturnal -diurnal anthesis allowed reliance on both bats and hummingbirds (Dar et al. 2006) . In 465 contrast, we consider it unlikely that H. pseudomelanostele or L. acanthurus generate meaningful 466 pollination interference for H. acranthus at our site, as both species occur at much lower densities, and L. 467 acanthurus shows little to no flowering overlap with H. acranthus. However, this is a possibility in several 468 other Lomas populations of coastal Peru, where this cactus coexists with multiple hummingbird- and bat-469 pollinated species, often showing reproductive compatibility even across genus (Arakaki et al. 2020). 470 Geographic variation in pollination systems has been documented in cacti. Schlumpberger et al. 471 (2009) found shifts from bee pollination (short, morning-opening, low-nectar flowers) sphingid pollination 472 (long-tubed, dusk-opening, nectar-rich flowers) across populations of Echinopsis ancistrophora along an 473 altitudinal gradient in Argentina. In Mexico, Stenocereus thurberi also showed spatial differences: northern 474 and central populations relied mainly on bats with no pollen limitation, whereas southern populations on 475 mixed pollinators and showed pollen limitation, likely due to temporal variation in pollinator availability 476 (Bustamante et al. 2010). Pachycereus pecten -aboriginum also displayed geographic differentiation in 477 anthesis and nectar secretion that aligns with nocturnal versus mixed nocturnal -diurnal pollination in 478 Valiente-Banuet et al. (2004). This tendency to more generalized pollination systems has also been 479 documented in other Pachycereeae such as Carnegia gigantea (Fleming et al., 1996 ); supporting that 480 generalized strategies appear to be more reliable in zones where bats predictability is lower (e.g., northern 481 Mexico versus central and Southern regions, Rojas-Martínez et al. 1999). These dynamics remain unstudied 482 in Peruvian cacti. The mixed pollination strategy we propose for H. acranthus may be shaped by local 483 factors such as habitat condition, pollinator abundance, and the presence or absence of co-flowering cacti. 484 However, since our data come s from a single site and one sampling year, it cannot reveal if pollination 485 mechanisms vary across space or time, or whether such variation explain s possible differences among 486 populations. Wider geographic coverage and long-term studies will be essential to address these questions. 487 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint

488 Our study shows that H. acranthus exhibits floral traits aligned with nocturnal bat pollination while 489 remaining compatible with diurnal hummingbird pollination, with both likely serving as the main effective 490 pollinators. In contrast, bees and sphingids mismatched key floral traits and likely acted primarily as nectar 491 or pollen thieves. We found no differences in floral traits or visitation frequencies between color morphs, 492 providing no support for our initial hypothesis. Overall, our findings indicate a mixed vertebrate pollination 493 system involving both bats and hummingbirds. In the strongly seasonal Lomas ecosystem, where pollinator 494 availability can shift across space and time, maintaining interactions with multiple pollinator guilds may 495 enhance reproductive stability. Our results highlight the need for broader spatial and temporal studies of H. 496 acranthus reproduction. 497

498 We thank A. Ceroni and M. Flores (Weberbauer Herbarium -MOL) for their support in obtaining 499 funding and for their contributions during the design of the study ; C. Reynel and S. Terreros (Forestry 500 Herbarium-MOLF) for receiving the botanical samples; M. Alvarado and E. Medina (Department of 501 Entomology-UNMSM) for receiving the arthropod samples and assisting with their taxonomic 502 identification. This research was funded by the ‘XI Concurso de fondos de investigación para círculos de 503 investigación 2021’ from La Molina National Agrarian University (Lima, Peru). This study complies with 504 legal requirements set by the Peruvian Ministry of Agriculture (MINAGRI) and the Forestry and Fauna 505 Service (SERFOR) for the collection of plant and insect specimens (RD N° D0000046 -2024-MIDAGRI-506 SERFOR-DGGSPFFS-DGSPF). 507 508 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint

509 Albuquerque-Lima S., Taylor N.P., Zappi D.C., Machado I.C. (2023) Floral specialization and bat pollination 510 in subtribe Cereinae (Cactaceae): a morphological approach. Diversity, 15, 207. https://doi.org/10.3390/d15020207 511 512 Arakaki M., Ostolaza C., Cáceres F., Roque J. (2006) Cactaceae endémicas del Perú. Revista Peruana de 513 Biología, 13, 193–219. 514 515 Arakaki M., Soltis D.E., Speranza P. (2007) New chromosome counts and evidence of polyploidy in 516 Haageocereus and related genera in tribe Trichocereeae and other tribes of Cactaceae. Brittonia, 59, 290 –297. 517 https://doi.org/10.1663/0007-196X(2007)59[290:NCCAEO]2.0.CO;2 518 519 Arakaki M., Speranza P., Soltis P.S., Soltis D.E. (2021) Examination of reticulate evolution involving 520 Haageocereus and Espostoa. Haseltonia, 27, 102–112. https://doi.org/10.2985/026.027.0112 521 522 Armbruster W.S. (2017) The specialization continuum in pollination systems: diversity of concepts and 523 implications for ecology, evolution and conservation. Functional ecology, 31, 88–100. https://doi.org/10.1111/1365-524 2435.12783 525 526 Armbruster W.S., Keller S., Matsuki M., Clausen T.P. (1989) Pollination of Dalechampia magnoliifolia 527 (Euphorbiaceae) by male euglossine bees. American Journal of Botany , 76, 1279 –1285. 528 https://doi.org/10.1002/j.1537-2197.1989.tb15109.x 529 530 Baker H.G. (1961) The adaptation of flowering plants to nocturnal and crepuscular pollinators. The Quarterly 531 Review of Biology, 36, 64–73. 532 533 Baker H.G. (1975) Sugar Concentrations in Nectars from Hummingbird Flowers. Biotropica, 7, 37 –41. 534 https://doi.org/10.2307/2989798 535 536 Ballesteros-Mejia L., Lima N.E., Lima -Ribeiro, M.S., Collevatti, R.G. (2016) Pollination mode and mating 537 system explain patterns in genetic differentiation in neotropical plants. PLoS One , 11, e0158660. 538 https://doi.org/10.1371/journal.pone.0184674 539 540 Barton K. (2024) MuMIn: Multi -Model Inference. R package v . 1.48.4, https://cran.r-541 project.org/package=MuMIn 542 543 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Brooks M.E., Kristensen K., Van Benthem K.J., Magnusson A., Berg C.W., Nielsen A., Skaug H.J., Mächler 544 M., Bolker B.M. (2017) glmmTMB Balances Speed and Flexibility Among Packages for Zero-inflated Generalized 545 Linear Mixed Modeling. The R journal, 9, 378–400. https://doi.org/10.32614/RJ-2017-066 546 547 Burkle L.A., Alarcón R. (2011) The future of plant –pollinator diversity: understanding interaction networks 548 across time, space, and global change. American journal of botany, 98, 528–538. https://doi.org/10.3732/ajb.1000391 549 550 Burnham K.P., Anderson D.R. (2002) Model Selection and Multimodel Inference: A Practical Information-551 Theoretic Approach. Springer New York, New York, US: 488 pp. https://doi.org/10.1007/b97636 552 553 Bustamante E., Casas A., Búrquez A. (2010) Geographic variation in reproductive success of Stenocereus 554 thurberi (Cactaceae): Effects of pollination timing and pollinator guild. American Journal of Botany, 97, 2020–2030. 555 https://doi.org/10.3732/ajb.1000071 556 557 Calderón N., Zappi D., Taylor N., Ceroni A. (2007) Taxonomy and conservation of Haageocereus Backeb. 558 (Cactaceae) in Peru. Bradleya, 25, 45–124. https://doi.org/10.25223/brad.n25.2007.a8 559 560 Cano A.C., Roque J., Arakaki M., Arana C., La Torre M., Llerena N., Refulio N. (1999) Diversidad florística 561 de las Lomas de Lachay (Lima) durante el evento "El Niño 1997-98". Revista Peruana de Biología, 6, 125–132. 562 563 Cardona J., Lara C., Ornelas J.F. (2020) Pollinator divergence and pollination isolation between hybrids with 564 different floral color and morphology in two sympatric Penstemon species. Scientific reports , 10, 8126. 565 https://doi.org/10.1038/s41598-020-64964-8 566 567 Ceroni A., Castro -Cepero V., Teixeira V., Redolfi I. (2007) Neoraimondia arequipensis subsp. roseiflora 568 (Werdermann & Backeberg) Ostolaza (Cactaceae): axis of the interactions in arid ecosystems. Ecología Aplicada, 6, 569 155–168. 570 571 Dafni A. (1992) Pollination Ecology: A Practical Approach. Oxford University Press, Oxford, UK: 250 pp. 572 573 Dar S., Arizmendi M.D.C., Valiente-Banuet A. (2006) Diurnal and nocturnal pollination of Marginatocereus 574 marginatus (Pachycereeae: Cactaceae) in Central Mexico. Annals of Botany , 97, 423 -427. 575 https://doi.org/10.1093/aob/mcj045 576 577 Dellinger A.S. (2020) Pollination syndromes in the 21st century: where do we stand and where may we 578 go?. New Phytologist, 228, 1193–1213. https://doi.org/10.1111/nph.16793 579 580 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Dourojeanni M.J. (2018) Áreas Naturales Protegidas del Perú: El Comienzo . Editora Grijley, Lima , Peru: 581 330 pp. 582 583 Faegri K., van der Pijl L. (1979) The principles of pollination ecology. Pergamon Press, Oxford, UK: 244 pp. 584 585 Fagua J.C., Ackerman J.D. (2011) Consequences of floral visits by ants and invasive honeybees to the 586 hummingbird‐pollinated, Caribbean cactus Melocactus intortus. Plant Species Biology , 26, 193–204. 587 https://doi.org/10.1111/j.1442-1984.2011.00319.x 588 589 Fenster C.B., Armbruster W.S., Wilson P., Dudash M.R., Thomson J.D. (2004) Pollination syndromes and 590 floral specialization. Annual Review of Ecology, Evolution and Systematics, 35, 375–403. 591 https://doi.org/10.1146/annurev.ecolsys.34.011802.132347 592 593 Ferreyra R. (1993) Registros de la vegetación en la costa peruana en relación con el fenómeno El 594 Niño. Bulletin de l’Institut français d’études andines, 22, 259–266. 595 596 Fleming T.H., Geiselman C., Kress, W.J. (2009) The evolution of bat pollination: a phylogenetic 597 perspective. Annals of botany, 104, 1017–1043 https://doi.org/10.1093/aob/mcp197 598 599 Fleming T.H., Tuttle M.D., Horner M.A. (1996) Pollination biology and the relative importance of nocturnal 600 and diurnal pollinators in three species of Sonoran Desert columnar cacti. The Southwestern Naturalist, 257-269. 601 602 Fleming T.H., Valiente-Banuet A. (Eds) (2002) Columnar cacti and their mutualists: evolution, ecology, and 603 conservation. University of Arizona Press, Arizona, US: 386 pp. 604 605 Fleming T.H., Sahley C.T., Holland J. N., Nason J.D., Hamrick J.L. (2001) Sonoran Desert columnar cacti 606 and the evolution of generalized pollination systems. Ecological Monographs , 71, 511–530. 607 https://doi.org/10.1890/0012-9615(2001)071[0511:SDCCAT]2.0.CO;2 608 609 Gonzalez-Terrazas T.P., Martel C., Milet-Pinheiro P., Ayasse M., Kalko E.K., Tschapka, M. (2016) Finding 610 flowers in the dark: nectar -feeding bats integrate olfaction and echolocation while foraging for nectar. Royal Society 611 Open Science, 3, 160199. https://doi.org/10.1098/rsos.160199 612 613 Grillo S., Arana C. (2016) Polinizadores y visitantes florales de Haageocereus acranthus y H. 614 pseudomelanostele (Cactaceae) en la Reserva Nacional de Lomas de Lachay, Perú. XXV Reunión Científica del 615 Instituto de Investigación de Ciencias Biológicas “Antonio Raimondi” - Universidad Nacional Mayor de San Marcos, 616 Lima, Peru. 617 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint 618 Grant V., Connell W.A. (1979) The association between Carpophilus beetles and cactus flowers. Plant 619 Systematics and Evolution, 133, 99–102. https://doi.org/10.1007/BF00985884 620 621 Grant V., Grant K.A. (1979) The pollination spectrum in the southwestern American cactus flora. Plant 622 Systematics and Evolution, 133, 29-37. https://doi.org/10.1007/BF00985877 623 624 Hartig F. (2018) DHARMa: Residual Diagnostics for Hierarchical (Multi-Level / Mixed) Regression Models. 625 R package version 0.4.7. http://florianhartig.github.io/DHARMa/ 626 627 Ibarra‐Cerdeña C.N., Iñiguez‐Dávalos L.I., Sánchez‐Cordero V. (2005) Pollination ecology of Stenocereus 628 queretaroensis (Cactaceae), a chiropterophilous columnar cactus, in a tropical dry forest of Mexico. American Journal 629 of Botany, 92, 503–509. https://doi.org/10.3732/ajb.92.3.503 630 631 Kay K. M., Anderson B. (2025) Beyond the Grant –Stebbins model: floral adaptive landscapes and plant 632 speciation. Annals of Botany, 136, 699-720. https://doi.org/10.1093/aob/mcaf096 633 634 Kay K.M., Sargent, R.D. (2009) The role of animal pollination in plant speciation: integrating ecology, 635 geography, and genetics. Annual Review of Ecology, Evolution, and Systematics , 40, 637 –656. 636 https://doi.org/10.1146/annurev.ecolsys.110308.120310 637 638 Lambert R. (2021) Examinando la relación entre la planeación y la urbanización periférica en Lima. Ensayo: 639 Revista de arquitectura, urbanismo y territorio, 2, 71–93. https://doi.org/10.18800/ensayo.202102.004 640 641 Lendel A. (2013) South American Cacti in time and space: studies on the diversification of the tribe Cereeae, 642 with particular focus on subtribe Trichocereinae (Cactaceae). University of Zürich, Zürich, Switzerland: 214 pp. 643 644 Lenth R.V. (2018) Least-Squares Means: The R Package lsmeans. Journal of Statistical Software, 69, 1–33. 645 https://doi.org/10.18637/jss.v069.i01 646 647 LeVan K.E., Hung K.L.J., McCann K.R., Ludka J.T., Holway D.A. (2014) Floral visitation by the Argentine 648 ant reduces pollinator visitation and seed set in the coast barrel cactus, Ferocactus viridescens. Oecologia, 174, 163–649 171. https://doi.org/10.1007/s00442-013-2739-z 650 651 Lüdecke D., Ben-Shachar M.S., Patil I., Waggoner P., Makowski D. (2021) performance: An R package for 652 Assessment, Comparison and Testing of Statistical Models. Journal of Open Source Software , 6. 653 https://doi.org/10.21105/joss.03139 654 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint 655 Maguiña R., Amanzo J. (2016) Diet and pollinator role of the long -snouted bat Platalina genovensium in 656 Lomas ecosystem of Peru. Tropical Conservation Science , 9, 1940082916674288 . 657 https://doi.org/10.1177/1940082916674288 658 659 Mandujano M.D.C., Carrillo -Angeles I., Martínez -Peralta C., Golubov J. (2010) Reproductive biology of 660 Cactaceae. In: Ramawat, K. (Eds), Desert plants: Biology and biotechnology. Springer, Berlin, Germany: 197–230. 661 662 Martins C., Oliveira R., Aguiar L.M., Antonini Y. (2020) Pollination biology of the endangered columnar 663 cactus Cipocereus crassisepalus: a case of close relationship between plant and pollinator. Acta Botanica 664 Brasilica, 34, 177–184. https://doi.org/10.1590/0102-33062019abb0219 665 666 McIntosh M.E. (2005) Pollination of two species of Ferocactus: interactions between cactus-specialist bees 667 and their host plants. Functional Ecology, 19, 727–734. https://www.jstor.org/stable/3599235 668 669 Medel R., López-Aliste M., Fontúrbel F.E. (2022) Hummingbird -plant interactions in Chile: An ecological 670 review of the available evidence. Avian Research, 13, 100051. https://doi.org/10.1016/j.avrs.2022.100051 671 672 Minnaar C., Anderson B., de Jager, M.L., Karron J.D. (2019) Plant–pollinator interactions along the pathway 673 to paternity. Annals of Botany, 123, 225–245. https://doi.org/10.1093/aob/mcy167 674 675 Miranda-Jácome A., Fernández-Tlapa F., Munguía-Rosas M.A. (2021) Visiting and feeding behavior of sap 676 beetles ( Carpophilus lugubris) in the flowers of a chiropterophilic columnar cactus ( Pilosocereus 677 leucocephalus). Journal of Arid Environments, 189, 104482. https://doi.org/10.1016/j.jaridenv.2021.104482 678 679 Miyake T., Yahara T. (1999) Theoretical evaluation of pollen transfer by nocturnal and diurnal pollinators: 680 when should a flower open?. Oikos, 86, 233–240. https://doi.org/10.2307/3546441 681 682 Moya-Méndez N. C. (2006) Haageocereus: Taxonomy for the Conservation of the Genus in Peru. The Open 683 University, Milton Keynes, UK: 164 pp. 684 685 Muchhala N., Thomson J.D. (2010) Fur versus feathers: pollen delivery by bats and hummingbirds and 686 consequences for pollen production. The American Naturalist, 175, 717–726. 687 688 Nassar J.M., Ramírez N., Linares O. (1997) Comparative pollination biology of Venezuelan columnar cacti 689 and the role of nectar‐feeding bats in their sexual reproduction. American Journal of Botany , 84, 918 –927. 690 https://doi.org/10.2307/2446282 691 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint 692 Nelson X.J. (2023) A road map of jumping spider behavior. The Journal of Arachnology , 51, 139 –154. 693 https://doi.org/10.1636/JoA-S-22-011 694 695 Nicolay C.W., Winter Y. (2006) Performance analysis as a tool for understanding the ecological morphology 696 of flower visiting bats. In : Zubaid A., McCracken G.F., Kunz T. H. (Eds), Functional and Evolutionary Ecology of 697 Bats. Oxford University Press, Oxford, UK: 131–144. 698 699 Novoa S., Ledesma K., Linares R. (2022) Use of Camera Traps to Monitor the Fauna Associated with a 700 Cactus Community in the Coastal Desert of Ica, Peru. Cactus and Succulent Journal , 94, 28 –33. 701 https://doi.org/10.2985/015.094.0105 702 703 Novoa S., Redolfi I., Ceroni A., Arellano C. (2005) El forrajeo de la hormiga Camponotus sp. en los botones 704 florales del cactus Neoraimondia arequipensis subsp. roseiflora (Werdermann & Backeberg) Ostolaza 705 (Cactaceae). Ecología aplicada, 4, 83–87. 706 707 Ollerton J., Alarcón R., Waser N.M., Price M.V., Watts S., Cranmer L., Hingston A., Peter C.I., Rotenberry 708 J. (2009) A global test of the pollination syndrome hypothesis. Annals of botany , 103, 1471–1480. 709 https://doi.org/10.1093/aob/mcp031 710 711 Olesen J.M., Jordano P. (2002) Geographic patterns in plant –pollinator mutualistic networks. Ecology, 83, 712 2416–2424. https://doi.org/10.1890/0012-9658(2002)083[2416:GPIPPM]2.0.CO;2 713 714 Opedal Ø. H. (2019) The evolvability of animal‐pollinated flowers: towards predicting adaptation to novel 715 pollinator communities. New Phytologist, 221, 1128–1135. https://doi.org/10.1111/nph.15403 716 717 Ossa G., Zamora H.T., Velazco P.M. (2020) Platalina genovensium (Chiroptera: 718 Phyllostomidae). Mammalian Species, 52, 105–113. https://doi.org/10.1093/mspecies/seaa008 719 720 Ostolaza C. (2014) Todos los cactus del Perú. Ministerio del Ambiente, Lima, Peru: 538 pp. 721 722 Ostolaza C. (1996) A closer look at the conservation status of Cacti in the vicinity of Lima, Peru. British 723 Cactus & Succulent Journal, 14, 158–174. https://www.jstor.org/stable/42793423 724 725 Pellón J.J., Mendoza J.L., Quispe -Hure O., Condo F., Williams M. (2021). Exotic cultivated plants in the 726 diet of the nectar-feeding bat Glossophaga soricina (Phyllostomidae: Glossophaginae) in the city of Lima, Peru. Acta 727 Chiropterologica, 23, 107–117. https://doi.org/10.3161/15081109ACC2021.23.1.009 728 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint 729 Pérez‐Barrales R., Arroyo J., Scott Armbruster W. (2007) Differences in pollinator faunas may generate 730 geographic differences in floral morphology and integration in Narcissus papyraceus (Amaryllidaceae). Oikos, 116, 731 1904–1918. https://doi.org/10.1111/j.0030-1299.2007.15994.x 732 733 Posit team (2024) RStudio: Integrated Development Environment for R. Posit Software, PBC, Boston, USA. 734 735 POWO (2024) Plants of the World Online. Royal Botanic Gardens, Kew, London, UK. 736 737 Proctor M., Yeo P., Lack A. (1996) The natural history of pollination. Timber Press, Oregon, USA: 479 pp. 738 739 R Core Team (2024) R: A Language and Environment for Statistical Computing. R Foundation for Statistical 740 Computing, Vienna, Austria. 741 742 Rico-Gray V. (1989) The importance of floral and circum -floral nectar to ants inhabiting dry tropical 743 lowlands. Biological Journal of the Linnean Society , 38, 173 –181. https://doi.org/10.1111/j.1095-744 8312.1989.tb01572.x 745 746 Rivera‐Marchand B., Ackerman J.D. (2006) Bat Pollination Breakdown in the Caribbean Columnar Cactus 747 Pilosocereus royenii. Biotropica, 38, 635–642. https://doi.org/10.1111/j.1744-7429.2006.00179.x 748 749 Rocha E.A., Domingos-Melo A., Zappi D.C., Machado I. C. (2019) Reproductive biology of columnar cacti: 750 are ba ts the only protagonists in the pollination of Pilosocereus, a typical chiropterophilous genus? Folia 751 Geobotanica, 54, 239–256. 752 753 Rojas‐Martínez A., Valiente‐Banuet A., Del Coro Arizmendi M., Alcántara‐Eguren A., Arita H.T. (1999) 754 Seasonal distribution of the long‐nosed bat (Leptonycteris curasoae) in North America: does a generalized migration 755 pattern really exist?. Journal of Biogeography, 26, 1065-1077. https://doi.org/10.1046/j.1365-2699.1999.00354.x 756 757 Rosas‐Guerrero V., Aguilar R., Martén‐Rodríguez S., Ashworth L., Lopezaraiza ‐Mikel M., Bastida J.M., 758 Quesada M. (2014) A quantitative review of pollination syndromes: do floral traits predict effective 759 pollinators?. Ecology letters, 17, 388–400. https://doi.org/10.1111/ele.12224 760 761 Rowley G. (1980) Pollination syndromes and cactus taxonomy. The Cactus and Succulent Journal of Great 762 Britain, 42, 95–98. 763 764 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Sahley C.T. (1996) Bat and hummingbird pollination of an autotetraploid columnar cactus, 765 Weberbauerocereus weberbaueri (Cactaceae). American Journal of Botany , 83, 1329 –1336. 766 https://doi.org/10.1002/j.1537-2197.1996.tb13918.x 767 768 Sahley C.T., Baraybar L. (1996) Natural history of the long -snouted bat, Platalina genovensium 769 (Phyllostomidae: Glossophaginae) in southwestern Peru. Vida Silvestre Neotropical, 5, 101–109. 770 771 [dataset] Servicio Nacional de Meteorología e Hidrología del Perú - SENAMHI; 2022; Estación: VILLA 772 MARIA DEL TRIUNFO; Datos Hidrometeorológicos; https://www.senamhi.gob.pe/?p=estaciones 773 774 Schlumpberger B.O. (2011) A survey on pollination modes in cacti and a potential key innovation. In: Patiny 775 S. (Ed), Evolution of Plant -Pollinator Relationships . Cambridge University Press , Cambridge, UK: 301 –319. 776 https://doi.org/10.1017/CBO9781139014113.011 777 778 Schlumpberger B.O., Cocucci A.A., Moré M., Sérsic A.N., Raguso R.A. (2009) Extreme variation in floral 779 characters and its consequences for pollinator attraction among populations of an Andean cactus. Annals of 780 Botany, 103, 1489–1500. https://doi.org/10.1093/aob/mcp075 781 782 Schneider C.A., Rasband W.S., Eliceiri K.W. (2012) NIH Image to ImageJ: 25 years of image 783 analysis. Nature methods, 9, 671–675. https://doi.org/10.1038/nmeth.2089 784 785 Silva W.R., Sazima M. (1995) Hawkmoth pollination in Cereus peruvianus, a columnar cactus from 786 southeastern Brazil. Flora - Morphology, Distribution, Functional Ecology of Plants , 190, 339 –343. 787 https://doi.org/10.1016/S0367-2530(17)30674-6 788 789 Smith C.E., Stevens J.T., Temeles E.J. , Ewald P.W., Hebert R.J., Bonkvsky R.L. (1996) Effect of floral 790 orifice width and shape on hummingbird -flower interactions. Oecologia, 106, 482–492. 791 https://doi.org/10.1007/BF00329706 792 793 Solís‐Montero L., Vallejo‐Marín M. (2017) Does the morphological fit between flowers and pollinators affect 794 pollen deposition? An experimental test in a buzz‐pollinated species with anther dimorphism. Ecology and 795 Evolution, 7, 2706–2715. https://doi.org/10.1002/ece3.2897 796 797 Stebbins G.L. (1970) Adaptive radiation of reproductive characteristics in angiosperms, I: Pollination 798 mechanisms. Annual Review of Ecology and Systematics, 1, 307–326. 799 800 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Su Q., Qi L., Zhang W., Yun Y., Zhao Y., Peng Y. (2020) Biodiversity survey of flower -visiting spiders 801 based on literature review and field study. Environmental Entomology , 49, 673 –682. 802 https://doi.org/10.1093/ee/nvaa022 803 804 Thompson J.N. (2002) The geographic mosaic of coevolution. University of Chicago Press, Chicago, USA: 805 400 pp. https://doi.org/10.7208/9780226118697 806 807 Thorp R.W. (2000) The collection of pollen by bees. In: Dafni A., Hesse M., Pacini E. (Eds), Pollen and 808 Pollination. Springer, Vienna, Austria: 211–233. https://doi.org/10.1007/978-3-7091-6306-1_11 809 810 Tovar, C., Sánchez Infantas, E., Teixeira Roth, V. (2018) Plant community dynamics of Lomas fog oasis of 811 Central Peru after the extreme precipitation caused by the 1997 -98 El Niño event. PLoS One , 13, e0190572. 812 https://doi.org/10.1371/journal.pone.0190572 813 814 Tremlett C.J., Moore M., Chapman M.A., Zamora‐Gutierrez V., Peh K.S.H. (2020) Pollination by bats 815 enhances both quality and yield of a major cash crop in Mexico. Journal of Applied Ecology, 57, 450 –459. 816 https://doi.org/10.1111/1365-2664.13545 817 818 Ulloa C., Zarucchi J.L., León B. (2004) Diez años de adiciones a la flora del Perú: 1993 -2003. Arnaldoa, 819 Trujillo, Peru: 242 pp. 820 821 Valiente‐Banuet A., Molina‐Freaner F., Torres A., Arizmendi M.D.C., Casas A. (2004) Geographic 822 differentiation in the pollination system of the columnar cactus Pachycereus pecten‐aboriginum . American Journal 823 of Botany, 91(6), 850-855. https://doi.org/10.3732/ajb.91.6.850 824 825 Valiente-Banuet A., Arizmendi M.D.C., Rojas -Martínez A., Domínguez -Canseco L. (1996) Ecological 826 relationships between columnar cacti and nectar -feeding bats in Mexico. Journal of Tropical Ecology , 12, 103–119. 827 https://doi.org/10.1017/S0266467400009330 828 829 Vamosi J.C., Vamosi S.M. (2010) Key innovations within a geographical context in flowering plants: 830 towards resolving Darwin’s abominable mystery. Ecology Letters , 13, 1270 –1279. https://doi.org/10.1111/j.1461-831 0248.2010.01521.x 832 833 Walter H.E. (2010) Floral biology of Echinopsis chiloensis ssp. chiloensis (Cactaceae): evidence for a mixed 834 pollination syndrome. Flora - Morphology, Distribution, Functional Ecology of Plants , 205, 757 –763. 835 https://doi.org/10.1016/j.flora.2009.12.038 836 837 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Westerkamp C.H. (1996) Pollen in bee‐flower relations some considerations on melittophily. Botanica 838 Acta, 109, 325–332. https://doi.org/10.1111/j.1438-8677.1996.tb00580.x 839 840 Wasserthal L.T. (2001) Anpassungen bei Sphingiden zur Vermeidung von Spinnen-und Fledermausattacken. 841 Verhandlungen Westdeustchen Entomologentag Dusseldorf, 2000, 13–30. 842 843 Wenzell K.E., Neequaye M., Paajanen P., Hill L., Brett P., Byers K. J. (2025) Within-species floral evolution 844 reveals convergence in adaptive walks during incipient pollinator shift. Nature Communications , 16, 2721. 845 https://doi.org/10.1038/s41467-025-57639-3 846 847 Winter Y., von Helversen O. (2003) Operational tongue length in phyllostomid nectar-feeding bats. Journal 848 of mammalogy, 84, 886–896. https://doi.org/10.1644/BWG-032 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Figures 875 Figure 1. (A) Geographic location and composition of the study area in Peru, showing the study area (SA) 876 within the Lomas ecosystem (LE), near the Lurin River (LR), and surrounded by agricultural and semi -877 urban land. (B) Individual of H. acranthus in the study area. 878 879 880 881 882 883 884 885 886 887 888 889 890 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Figure 2. (A) Longitudinal cut of H. acranthus flowers indicating measured morphometric variables. (B) 891 Size and shape variation in flowers of both morphotypes. Frontal view showing perianth color and 892 reproductive structures of (C) white and (D) pink-red flower morphotypes. 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Figure 3. Monthly mean counts of (A) floral buds and flowers, and (B) unripe and ripe fruits observed 908 throughout 2022. Vertical bars indicate the standard error of the mean. Shaded gray areas correspond to the 909 humid Lomas season. 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Figure 4. Temporal variation in (A) flower aperture, (B) nectar volume, (C) nectar sucrose content, and 925 (D) nectar energy production as a function of time after midday. 926 Solid lines represent model predictions, and dotted and dashed lines indicate 95% confidence intervals. In 927 panel (A), the model includes the effect of morphotype, providing separate estimates and confidence 928 intervals for each form. Shaded gray areas correspond to the nocturnal period. 929 930 931 932 933 934 935 936 937 938 939 940 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Figure 5. Floral visitors of H. acranthus: (A–B) Rhodopis vesper (Trochilidae); ( C–D) Platalina 941 genovenisum (Phyllostomidae); (E–F) Manduca sp. (Sphingidae). 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Figure 6. Scaled density of floral visits of Rhodopis vesper, Platalina genovensium, and sphingid moths 958 over time for (A) white and (B) pink-red flower morphotypes. Ring plots indicate the percentage of total 959 visits contributed by each visitor group for each morphotype. Shaded gray regions indicate the nocturnal 960 period. Shaded gray areas correspond to the nocturnal period. 961 962 963 964 965 966 967 968 969 970 971 972 973 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Tables 974 Table 1. Summary of the best -fitting models for each response variable: (A) flower aperture, (B) nectar 975 volume, (C) nectar sugar concentration, (D) nectar energy content, and (E) visitation frequency. Fixed 976 effects are expressed as estimates with their standard errors (SE), Z-values (Z), and p-values (P). Estimates 977 are relative to white flowers for morphotype and to sphingid moths (Manduca sp.) for floral visitors. 978 Response variable Fixed effect Estimate SE Z P (A) Flower aperture Intercept 4.11 0.17 24.88 <0.001 T -0.97 1.59 -0.61 0.540 T2 -18.83 1.58 -11.92 <0.001 T3 4.34 1.54 2.82 0.005 T : Morph red-pink -4.32 2.26 -1.91 0.056 T2 : Morph red-pink 4.89 2.24 2.18 0.029 T3 : Morph red-pink -2.61 2.25 -1.16 0.247 (B) Nectar volume Intercept 3.33 0.15 22.45 <0.001 T -0.11 2.05 -0.05 0.957 T2 -9.28 1.76 -5.29 <0.001 T3 8.14 1.81 4.51 <0.001 (C) Nectar sugar content Intercept 20.39 0.43 47.27 <0.001 T -13.77 1.60 -8.58 <0.001 (D) Nectar energy Intercept 4.33 0.22 19.69 <0.001 T -1.19 2.55 -0.47 0.639 T2 -7.71 2.06 -3.74 <0.001 T3 8.54 1.79 4.78 <0.001 (E) Visitation frequency Intercept -5.63 0.72 -7.83 <0.001 Visitor P. genovenisum 1.20 0.47 2.59 0.01 Visitor R. vesper 2.59 0.42 6.12 <0.001 ´:´ indicates an interaction term. ´T´ represents time after midday (in hours). 979 980 981 982 983 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint Table 2. Summary of registered floral visitors, including visitor group, taxonomic classification, visit type, 984 floral resource utilized, and activity period. 985 Visitor group Order Family Species Visit type Floral resource Activity period Bees Hymenoptera Apidae Apis mellifera I/L Pollen Diurnal Halictidae - I Pollen Diurnal Ants Hymenoptera Formicidae Solenopsis sp. I Nectar Diurnal/Nocturnal Formicidae Linepithema sp. I Nectar Diurnal/Nocturnal Flies Diptera Phoridae - I Pollen Diurnal Syrphidae - I Pollen Diurnal Beetles Coleoptera Nitidulidae Carpophilus sp. I Pollen, plant tissue Diurnal/Nocturnal Spiders Araneae Salticidae - I Hunting ground Diurnal/Nocturnal Anyphaenidae - I Hunting ground Diurnal/Nocturnal Thomisidae - I Hunting ground Diurnal/Nocturnal Sphingids Lepidoptera Sphingidae Manduca sp. I/L Nectar Nocturnal Hummingbirds Apodiformes Trochillidae Rhodopis vesper L Nectar Diurnal Bats Chiroptera Phyllostomidae Platalina genovensium L Nectar Nocturnal ´I´ and ´L´ represent illegitimate and legitimate visits to flowers, respectively. 986 987 988 989 990 991 992 993 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 9, 2026. ; https://doi.org/10.64898/2026.01.08.698271doi: bioRxiv preprint