Philippa Kaur

IPCC. Assessment Report 6 Climate Change 2021: The Physical Science Basis. (2021).Angilletta, M. J. Thermal adaptation: a theoretical and empirical synthesis. Oxford University Press (Elsevier, 2009).Sunday, J. M. et al. Thermal-safety margins and the necessity of thermoregulatory behavior across latitude and elevation. Proc. Natl Acad. Sci. USA 111, 5610–5615 (2014).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Hampe, A. & Petit, R. J. Conserving biodiversity under climate change: the rear edge matters. Ecol. Lett. 8, 461–467 (2005).PubMed 

Google Scholar 
Parmesan, C. Ecological and evolutionary responses to recent climate change. Annu. Rev. Ecol. Evol. Syst. 37, 637–669 (2006).
Google Scholar 
Ma, C. S., Ma, G. & Pincebourde, S. Survive a warming climate: insect responses to extreme high temperatures. Annu. Rev. Entomol. 66, 163–184 (2021).CAS 
PubMed 

Google Scholar 
Hoffmann, A. A., Sørensen, J. G. & Loeschcke, V. Adaptation of Drosophila to temperature extremes: Bringing together quantitative and molecular approaches. J. Therm. Biol. 28, 175–216 (2003).
Google Scholar 
Oostra, V., Saastamoinen, M., Zwaan, B. J. & Wheat, C. W. Strong phenotypic plasticity limits potential for evolutionary responses to climate change. Nat. Commun. 9, 1005 (2018).Štětina, T., Koštál, V. & Korbelová, J. The role of inducible Hsp70, and other heat shock proteins, in adaptive complex of cold tolerance of the fruit fly (Drosophila melanogaster). PLoS One 10, 1–22 (2015).
Google Scholar 
Overgaard, J., Sørensen, J. G., Petersen, S. O., Loeschcke, V. & Holmstrup, M. Changes in membrane lipid composition following rapid cold hardening in Drosophila melanogaster. J. Insect Physiol. 51, 1173–1182 (2005).CAS 
PubMed 

Google Scholar 
Laland, K. N. et al. The extended evolutionary synthesis: Its structure, assumptions and predictions. Proc. R. Soc. B Biol. Sci. 282, 20151019 (2015).Sánchez-Bayo, F. & Wyckhuys, K. A. G. Worldwide decline of the entomofauna: A review of its drivers. Biol. Conserv. 232, 8–27 (2019).
Google Scholar 
Deutsch, C. A. et al. Increase in crop losses to insect pests in a warming climate. Science 361, 916–919 (2018).ADS 
CAS 
PubMed 

Google Scholar 
Sgrò, C. M., Terblanche, J. S. & Hoffmann, A. A. What can plasticity contribute to insect responses to climate change? Annu. Rev. Entomol. 61, 433–451 (2016).PubMed 

Google Scholar 
Sørensen, J. G., Kristensen, T. N. & Overgaard, J. Evolutionary and ecological patterns of thermal acclimation capacity in Drosophila: is it important for keeping up with climate change? Curr. Opin. Insect Sci. 17, 98–104 (2016).PubMed 

Google Scholar 
Gunderson, A. R. & Stillman, J. H. Plasticity in thermal tolerance has limited potential to buffer ectotherms from global warming. Proc. R. Soc. B Biol. Sci. 282, 20150401 (2015).Rohr, J. R. et al. The complex drivers of thermal acclimation and breadth in ectotherms. Ecol. Lett. 21, 1425–1439 (2018).PubMed 

Google Scholar 
Gunderson, A. R., Dillon, M. E. & Stillman, J. H. Estimating the benefits of plasticity in ectotherm heat tolerance under natural thermal variability. Funct. Ecol. 31, 1529–1539 (2017).
Google Scholar 
Barley, J. M. et al. Limited plasticity in thermally tolerant ectotherm populations: Evidence for a trade-off. Proc. R. Soc. B Biol. Sci. 288, 20210765 (2021).
Google Scholar 
Morley, S. A., Peck, L. S., Sunday, J. M., Heiser, S. & Bates, A. E. Physiological acclimation and persistence of ectothermic species under extreme heat events. Glob. Ecol. Biogeogr. 28, 1018–1037 (2019).
Google Scholar 
Kellermann, V. & van Heerwaarden, B. Terrestrial insects and climate change: adaptive responses in key traits. Physiol. Entomol. 44, 99–115 (2019).
Google Scholar 
Seebacher, F., White, C. R. & Franklin, C. E. Physiological plasticity increases resilience of ectothermic animals to climate change. Nat. Clim. Chang. 5, 61–66 (2015).ADS 

Google Scholar 
Pincebourde, S. & Woods, H. A. There is plenty of room at the bottom: microclimates drive insect vulnerability to climate change. Curr. Opin. Insect Sci. 41, 63–70 (2020).PubMed 

Google Scholar 
van Heerwaarden, B. & Kellermann, V. Does plasticity trade off with basal heat tolerance? Trends Ecol. Evol. 35, 874–885 (2020).PubMed 

Google Scholar 
Stevenson, R. D. The relative importance of behavioral and physiological adjustments controlling body temperature in terrestrial ectotherms. Am. Nat. 126, 362–386 (1985).
Google Scholar 
Donelson, J. M., Salinas, S., Munday, P. L. & Shama, L. N. S. Transgenerational plasticity and climate change experiments: Where do we go from here? Glob. Chang. Biol. 24, 13–34 (2018).ADS 
PubMed 

Google Scholar 
Kristensen, T. N. et al. Costs and benefits of cold acclimation in field-released Drosophila. Proc. Natl Acad. Sci. 105, 216–221 (2008).ADS 
CAS 
PubMed 

Google Scholar 
Bozinovic, F., Calosi, P. & Spicer, J. I. Physiological correlates of geographic range in animals. Annu. Rev. Ecol. Evol. Syst. 42, 155–179 (2011).
Google Scholar 
Chown, S. L., Gaston, K. J. & Robinson, D. Macrophysiology: large-scale patterns in physiological traits and their ecological implications. Funct. Ecol. 18, 159–167 (2004).
Google Scholar 
Overgaard, J., Hoffmann, A. A. & Kristensen, T. N. Assessing population and environmental effects on thermal resistance in Drosophila melanogaster using ecologically relevant assays. J. Therm. Biol. 36, 409–416 (2011).
Google Scholar 
Sgrò, C. M. et al. A comprehensive assessment of geographic variation in heat tolerance and hardening capacity in populations of Drosophila melanogaster from Eastern Australia. J. Evol. Biol. 23, 2484–2493 (2010).PubMed 

Google Scholar 
Kingsolver, J. G. & Huey, R. B. Size, temperature, and fitness: three rules. Evol. Ecol. Res. 10, 251–268 (2008).
Google Scholar 
Brown, J., Gillooly, J. F., Allen, A. P., Savage, V. M. & West, G. Toward a metabolic theory of ecology. Ecology 85, 1771–1789 (2004).
Google Scholar 
Stillwell, R. C., Blanckenhorn, W. U., Teder, T., Davidowitz, G. & Fox, C. W. Sex differences in phenotypic plasticity affect variation in sexual size dimorphism in insects: from physiology to evolution. Annu. Rev. Entomol. 55, 227 (2010).CAS 
PubMed 
PubMed Central 

Google Scholar 
Tarka, M., Guenther, A., Niemelä, P. T., Nakagawa, S. & Noble, D. W. A. Sex differences in life history, behavior, and physiology along a slow-fast continuum: a meta-analysis. Behav. Ecol. Sociobiol. 72, 1–13 (2018).
Google Scholar 
Pottier, P., Burke, S., Drobniak, S. M., Lagisz, M. & Nakagawa, S. Sexual (in)equality? A meta-analysis of sex differences in thermal acclimation capacity across ectotherms. Funct. Ecol. 35, 2663–2678 (2021).
Google Scholar 
Bowler, K. & Terblanche, J. S. Insect thermal tolerance: what is the role of ontogeny, ageing and senescence? Biol. Rev. 83, 339–355 (2008).PubMed 

Google Scholar 
Fawcett, T. W. & Frankenhuis, W. E. Adaptive explanations for sensitive windows in development. Front. Zool. 12, 1–14 (2015).
Google Scholar 
English, S. & Barreaux, A. M. The evolution of sensitive periods in development: insights from insects. Curr. Opin. Behav. Sci. 36, 71–78 (2020).
Google Scholar 
Overgaard, J., Kristensen, T. N. & Sørensen, J. G. Validity of thermal ramping assays used to assess thermal tolerance in arthropods. PLoS One 7, e32758 (2012).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Bak, C. W. et al. Comparison of static and dynamic assays when quantifying thermal plasticity of drosophilids. Insects 11, 1–11 (2020).
Google Scholar 
Rodrigues, Y. K. & Beldade, P. Thermal plasticity in insects’ response to climate change and to multifactorial environments. Front. Ecol. Evol. 8, 271 (2020).
Google Scholar 
Terblanche, J. S. & Hoffmann, A. Validating measurements of acclimation for climate change adaptation. Curr. Opin. Insect Sci. 41, 7–16 (2020).PubMed 

Google Scholar 
Loeschcke, V. & Hoffmann, A. A. The detrimental acclimation hypothesis. Trends Ecol. Evol. 17, 407–408 (2002).
Google Scholar 
Cossins, A. R. & Bowler, K. Temperature Biology of Animals. (Chapman and Hall, 1987).Pintor, A. F. V., Schwarzkopf, L. & Krockenberger, A. K. Extensive acclimation in ectotherms conceals interspecific variation in thermal tolerance limits. PLoS One 11, e0150408 (2016).PubMed 
PubMed Central 

Google Scholar 
Rezende, E. L., Tejedo, M. & Santos, M. Estimating the adaptive potential of critical thermal limits: methodological problems and evolutionary implications. Funct. Ecol. 25, 111–121 (2011).
Google Scholar 
Allen, J. L., Chown, S. L., Janion-Scheepers, C. & Clusella-Trullas, S. Interactions between rates of temperature change and acclimation affect latitudinal patterns of warming tolerance. Conserv. Physiol. 4, cow053 (2016).Lutterschmidt, W. I. & Hutchison, V. H. The critical thermal maximum: History and critique. Can. J. Zool. 75, 1561–1574 (1997).
Google Scholar 
Terblanche, J. S. et al. Phenotypic plasticity and geographic variation in thermal tolerance and water loss of the tsetse Glossina pallidipes (Diptera: Glossinidae): Implications for distribution modelling. Am. J. Trop. Med. Hyg. 74, 786–794 (2006).PubMed 

Google Scholar 
Koricheva, J., Gurevitch, J. & Mengersen, K. Handbook of meta-analysis in ecology and evolution. Handbook of Meta-analysis in Ecology and Evolution (Princeton University Press, 2013).Suggitt, A. J. et al. Habitat microclimates drive fine-scale variation in extreme temperatures. Oikos 120, 1–8 (2011).
Google Scholar 
Oyen, K. J. & Dillon, M. E. Critical thermal limits of bumblebees (Bombus impatiens) are marked by stereotypical behaviors and are unchanged by acclimation, age or feeding status. J. Exp. Biol. 221, jeb165589 (2018).Bennett, J. M. et al. The evolution of critical thermal limits of life on Earth. Nat. Commun. 12, 1–9 (2021).
Google Scholar 
Bowler, K. Heat death in poikilotherms: Is there a common cause? J. Therm. Biol. 76, 77–79 (2018).PubMed 

Google Scholar 
MacMillan, H. A. & Sinclair, B. J. Mechanisms underlying insect chill-coma. J. Insect Physiol. 57, 12–20 (2011).CAS 
PubMed 

Google Scholar 
Hoffmann, A. A., Chown, S. L. & Clusella-Trullas, S. Upper thermal limits in terrestrial ectotherms: How constrained are they? Funct. Ecol. 27, 934–949 (2013).
Google Scholar 
Sandblom, E. et al. Physiological constraints to climate warming in fish follow principles of plastic floors and concrete ceilings. Nat. Commun. 7, 1–8 (2016).
Google Scholar 
Maclean, H. J. et al. Evolution and plasticity of thermal performance: An analysis of variation in thermal tolerance and fitness in 22 Drosophila species. Philos. Trans. R. Soc. B Biol. Sci. 374, 20180548 (2019).Addo-Bediako, A., Chown, S. L. & Gaston, K. J. Thermal tolerance, climatic variability and latitude. Proc. R. Soc. B Biol. Sci. 267, 739–745 (2000).CAS 

Google Scholar 
Sales, K. et al. Experimental heatwaves compromise sperm function and cause transgenerational damage in a model insect. Nat. Commun. 9, 1–11 (2018).ADS 
CAS 

Google Scholar 
Walsh, B. S. et al. Integrated approaches to studying male and female thermal fertility limits. Trends Ecol. Evol. 34, 492–493 (2019).PubMed 

Google Scholar 
Moher, D. et al. Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement. PLoS Med. 6, e1000097 (2009).Hadley, N. F. Water relations of terrestrial arthropods. (Academic Press, 1994).Hinchliff, C. E. et al. Synthesis of phylogeny and taxonomy into a comprehensive tree of life. PNAS. 112, 12764–12769 (2015).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Viechtbauer, W. Conducting meta-analyses in R with the metafor. J. Stat. Softw. 36, 1–48 (2010).
Google Scholar 
Barton, K. MuMIn: Multi-Model Inference. (2020).Nakagawa, S. et al. Methods for testing publication bias in ecological and evolutionary meta-analyses. Methods Ecol. Evol. 13, 4–21 (2022).
Google Scholar 
Macartney, E. L., Crean, A. J., Nakagawa, S. & Bonduriansky, R. Effects of nutrient limitation on sperm and seminal fluid: a systematic review and meta-analysis. Biol. Rev. 94, 1722–1739 (2019).PubMed 

Google Scholar  More

Data sources and pre-processingEach of the predictor variables used in our analysis (Table 1), as well as the dependent variable (fire hotspots) underwent pre-processing to transform the data into a format suitable to be passed to our CNN model for prediction. Here we briefly outline these processes and describe the method of generating a training and validation data set for model development. For further details about each predictor variable pre-processing, see Horton et al. (2021).Table 1 Model input data sources, citation, original resolution, and date ranges.Full size tableFire hotspotsWe used both Moderate Resolution Imaging Spectroradiometer (MODIS) and Visible Infrared Imaging Radiometer Suite (VIIRS) fire hotspot data as the dependent variable for use in our model development. As fire hotspots do not give precise locations, but rather indicate that a fire hotspot occurred within a grid cell of the size of the dataset (MODIS 1 km, VIIRS 375 m), we represented each fire hotspot as a 500 m buffered area around the centre point of each grid square identified. We used all fire hotspot occurrences with a confidence rating >50%.LandcoverWe use a collection of historic land cover maps generated by the Ministry of Forestry Indonesia from 1996 to 2016 at 2–3 year intervals38. Before use, we re-designated the land cover map classifications to reduce the number from 25 to just 8 (supplementary Table S2), which are ‘Primary and secondary dry forest’, ‘Swamp forest, ‘Swamp scrubland’, ‘Scrubland, Transition, and bare land’, ‘Riceland’, ‘Plantation’, ‘Settlements’, ‘water, and Cloud’.In addition to these 8 land cover classifications, we also derived a forest clearance index, which identifies areas cleared of forest and assigns an index value that is large negative (−10) immediately after clearing and degrades back towards 0 as time since clearing increases yearly. Areas that are re-forested are assigned large positive values (10) that degrade towards 0 yearly as time since afforestation increase25.Vegetation indicesAll vegetation indices were taken as pre-fire season 3-month averages from May to July. In addition to the original MODIS ET, PET, NDVI, and EVI products, we also included ‘normalised’ variables, whereby each vegetation index was expressed as the ratio of the same index taken at a reference site. The reference site was an area of dense primary forest outside of the EMRP area.Proximity to anthropogenic factorsThe distance to roads and settlement rasters were derived from OpenStreetMap data as the Euclidean distance to nearest feature in 250 m resolution. The same was done for all water bodies, which were then classified by hand into either canals or rivers. These features are taken as those shown in 2015 for all years, and therefore may misrepresent earlier years. However, the majority of canal development in the region took place between 1996 and 1998 and so should not differ dramatically from this date onwards.Oceanic Niño Index (ONI)We use a single value for the entire study area taken as the three-month average for the early fire season each year (July–September).Number of cloud daysUsing the state_1km band in the daily MODIS terra product (MOD09GA version 6), which classifies each pixel as either ‘no cloud’, ‘cloud’, ‘mixed’, or ‘unknown’, we counted the number of ‘cloud’ or ‘mixed’ designations for each pixel for the pre-fire season period May–July.Cross year normalisationAll predictor variables are normalised to be represented between 0 and 1 as the range between the minimum and maximum values for each variable that occur across all years, such that:$${V}_{{{{{{rm{norm}}}}}}}=frac{V-{V}_{{min }}}{{V}_{{max }}-{V}_{{min }}}$$where ({V}_{{{{{{rm{norm}}}}}}}) is the normalised version of the predictor variable (V), ({V}_{{max }}) is the maximum value within the training dataset across all years (2002–2019), and ({V}_{{min }}) is the minimum value within the training dataset across all years.Training and validation dataset assemblyOnce pre-processed, all predictor variable rasters were resampled to the same dimensions (with a resolution of 0.002 degrees in the WGS84 co-ordinate system) and stacked yearly, so that each year (2002–2019) comprised of a 31 feature maps input as a raster stack, with each feature map representing a different predictor variable. Each yearly stack was then split into tiles matching the input dimensions of the CNN model. Our final model was built to take an input size of 32 × 32 pixels (raster cells). Therefore, each yearly raster stack was split into many 32 × 32 × 31 raster stack tiles that span the defined study area. These were then converted to 3D arrays holding the values of all predictor variables for each raster stack tile.The same process was repeated for the yearly fire hotspot rasters used as the dependent variable in building our model. Each year was split into 32 × 32 × 1 tiles across the study area, and then converted to 3D arrays, each of which pairs with one predictor variable array.The 3D predictor variable arrays (dimensions: 32 × 32 × 31) were then stacked into one large 4D array containing all these individual tiles across all years (dimensions: W × 32 × 32 × 31, where W is a large value). The same was done with the 3D dependent variable arrays (dimension: 32 × 32 × 1), preserving the order so that each element in this large 4D array (dimensions: W × 32 × 32 × 1) matches with its counterpart in the predictor variable array.The order of this large 4D training data array was then randomised along the first dimension to avoid bias in passing to the CNN training algorithm, but the randomised re-ordering was repeated with the dependent variable array so as to preserve the elementwise pairing for cross-validation.Model development and applicationFire prediction requires the combination of spatial and temporal indicators to generate a probabilistic output for each location within a given study area. There is a need to preserve a certain level of proximity information, as the location of variables in relation to one another may have a substantial impact on the results. For example, a patch of secondary forest that is immediately adjacent to an area recently deforested may have a significantly higher probability of fire occurrence than an area surrounded entirely by primary forest.CNNs retain spatial features by employing a moving window of reference, known as a kernel, over the input image that captures these proximity relationships within the model structure. For this reason, CNNs are often used for image classification problems, and is an ideal model configuration for the problem of fire prediction across an area. Therefore, we have developed a CNN binary classification model using the Keras API package39 that builds on the TensorFlow machine learning platform40.Model structureCNN models typically apply a combination of kernel layers and dense layers that perform a series of transformations on the multi-channel input to either reduce it down to a single value, or to output an image the same width and height as the input with a single channel. These classification models can either assign a single value (binary classifier), or return one of many possible classifications.Kernels act on a subsection of the input stack (31 feature maps), assigning weights according to each cell’s position within the subsection to transform and combine the values into a new format to pass forward. As the kernel is applied to all subsections of the input stack, it transforms them to the new format, and builds a reconstituted image with dimensions that usually differ from the input. A dense layer will do the same operation, but acting only on a single grid cell of the input stack, acting at the same location upon all input feature maps within the stack at a time—using all values at that location (i.e., the 1 × 1 subsection) and transforming them according to assigned weights to pass forward a new set of channels to a single grid cell on the output stack. Each layer, either kernel or dense, may expand or contract the number of channels it passes forward. A kernel layer may also change the width and height dimensions of the subsection it passes forwards.We require an output that corresponds to a map of fire-occurrences; therefore our model needs to perform a series of transforms that preserve the width and height of the input, but reduce it to a single channel. The single channel in the output then represents the probability of each cell being classified as fire or not-fire (0–1).Our CNN model is comprised of 5 kernel layers (K1–K5 in Fig. 5), each acts on a 3 × 3 subsection and preserves width and height, passing forwards a transformed 3 × 3 section. Kernel K1 takes an input of 31 channels (predictor variables) but passes forward 128 channels to form the transformation T1 (Fig. 6). Kernels K2–K4 take inputs of 128 channels and pass forward 128 channels (T2–T4). Kernel K5 takes an input of 128 channels but passes forward 1 channel—the output. After each kernel applies its weights, there is an activation function applied before the values are passed on, which modify the answer to fit the necessary criteria to be a valid input to the next process. Kernels K1–K4 have a rectified linear (relu) activation function, which returns the input value if positive, and 0 if negative. Kernel K5 has a sigmoid activation function, that transforms the input values to between 0 and 1 such that negative values are transformed to 0.5.Fig. 6: Model structural diagram.Model structural diagram showing the input, 3 × 3 kernel layers (K1–K5), each transformation passed forwards (T1–T4) and the output, with all dimensions labelled.Full size imageModel training and validationWe used a stochastic gradient descent optimising function called Adam41 combined with a binary cross-entropy loss function to train the model against our fire-hotspot dataset iterated over 20 epochs. We split the data 70/30, using 70% as training data and 30% as validation data, recording accuracy, precision, and recall as the performance metrics, as well as the loss function itself.After model training, we applied the model to each yearly raster stack and compared the output against the fire-hotspot data for further model validation. Before validating the model outputs, we applied a simple 3 × 3 moving average window as a smoothing function to reduce the edge effects of tiling that are a by-product of having to split the study area into smaller tiles (32 × 32) for passing to the model. For this yearly validation, we again used the metrics accuracy, precision, and recall, such that:$${{{{{rm{Accuracy}}}}}}=100({{{{{rm{TP}}}}}}+{{{{{rm{TN}}}}}})/({{{{{rm{TP}}}}}}+{{{{{rm{TN}}}}}}+{{{{{rm{FP}}}}}}+{{{{{rm{FN}}}}}})$$$${{{{{rm{Precision}}}}}}=100({{{{{rm{TP}}}}}})/({{{{{rm{TP}}}}}}+{{{{{rm{FP}}}}}})$$$${{{{{rm{Recall}}}}}}=100({{{{{rm{TP}}}}}})/({{{{{rm{TP}}}}}}+{{{{{rm{FN}}}}}})$$where TP is true positive, TN is true negative, FP is false positive, and FN is false negative. These comparisons were made on a raster cell to raster cell basis after designating a 500 m buffer around each fire hotspot observation (MODIS and VIIRS data) and converting the buffers to a raster image of the same resolution and extent as the model prediction.ScenariosAfter validating the model performance, we built future scenarios to investigate the impact on fire occurrence of managing key anthropogenic features of the landscape: canals and land cover (Table 2).Table 2 Future scenario types and descriptions.Full size tableStudies have shown that unmanaged areas of heavily degraded or cleared swamp-forest are most susceptible to fires16,17,25,26,33,42. Therefore, we have built scenarios that investigate the possible impact of managing these areas by altering the model inputs to re-assign the land-cover designations ‘Swamp shrubland’ and ‘Scrubland’, as well as other land designation alterations. The first such restoration scenario investigates the impact of reforesting these areas by re-assigning the designations to ‘Swamp forest’. The second such scenario investigates the impact of converting these unmanaged areas to plantations by re-assigning the designations to ‘Plantation’. We also built two further land cover scenarios to investigate the impact of continued deforestation in the region by re-assigning the ‘Swamp forest’ designation to ‘Swamp shrubland’ and ‘Plantation’.We then built a scenario to investigate the impact of canal blocking on fire occurrence, modifying the proximity to canals model input by reducing the number of canals included in our proximity analysis to just two major canals, one that runs north-south, and one that runs west-east (Fig. 1). These canals could not practically be blocked due to their size and importance as navigation conduits.The final scenario simulates the combined impact of both re-foresting unmanaged degraded and cleared forest areas and the blocking of canals simultaneously.To evaluate the impact of each scenario on fire occurrences, we calculated the ratio of model predictions >0.5 probability (i.e., that a fire would occur in that raster cell) for each year for each scenario against the same year for the baseline scenario.Model use as a predictive toolTo evaluate the model’s potential to predict future fire distribution across the wider ex-Mega Rice Project area, we trained a second version of the model following the same methodology outlined above, but included only data from 2002 to 2018 in the training and test data passed to the model fitting algorithm. We then applied the model to the predictor variables corresponding to 2019 and compared model outputs to the observations of fire-occurrences by again looking at the metrics accuracy, precision, and recall. We also present a visual comparison of the outputs from the full model (2019 included in training data), the predictive model (2019 not included), and the observation data (MODIS and VIIRS hotspots). More

Study area and soil propertiesA field experiment was established in May 2012 at Shenyang Agro-Ecological Station (41°31′N, 123°22′E) of the Institute of Applied Ecology, Chinese Academy of Sciences, Northeast China. This region has a warm-temperate continental monsoon climate. The mean annual air temperature and annual precipitation are 7.5 °C and 680 mm, respectively. The soil is classified as Luvisol (FAO classification). The soil properties of the topsoil layer (0–20 cm) at the start of the experiment are as follows: SOC = 9.0 g kg−1, available NH4+–N = 1.18 mg kg−1; available NO3−–N = 9.04 mg kg−1; Olsen-P = 38.50 mg kg−1, available K = 97.90 mg kg−1, bulk density = 1.25 g cm−3, and pH = 5.8. The determination method of soil was shown in “Soil analysis” section.Field experimentThree treatments were established in this experiment: (1) mineral fertilizers (NPK); (2) pig manure incorporation at a local conventional AM application rate of 15 Mg ha−1 yr−1 (NPKM, 126 kg N ha−1 on dry weight); and (3) NPKM plus DMPP (3,4-Dimethylpyrazole phosphate) incorporation at a rate of 0.5% of applied urea (2.39 kg ha−1, 220 kg N/the N content of urea (0.46) × 0.5%) (NPKI + M). The treatments were applied following a randomized design across three replicate field plots (4 m × 5 m). Plots of different treatments remained unchanged in the same locations for 4 years. Each year, the composted pig manure (213 g C kg−1 and 22 g N kg−1 based on dry weight on average, characteristics of pig manure was listed in Table S1) was broadcasted evenly onto the plots a few days before maize planting, and ploughed to a depth of 20 cm by machine (TG4, Huaxing, China). For the respective treatments, urea (220 kg N ha−1 yr−1), calcium superphosphate (110 kg P2O5 ha−1 yr−1), and potassium chloride (110 kg K2O ha−1 yr−1) were applied on the same day as maize (Zea mays L.) was planted. The urea and inhibitor were fully mixed before application.Maize (cultivar was Fuyou #9) was planted on 3rd May 2012, 3rd May 2013, 6th May 2014, and 10th May 2015, at a spacing of 37 cm and 60 cm between rows. No irrigation was applied throughout the experimental period. Maize was harvested on 13th September 2012, 29th September 2013, 29th September 2014, and 29th September 2015, respectively. At harvest, maize yield and aboveground biomass yield were measured by harvesting all plants (20 m2) in each plot. The straw and grain were removed after each harvest and the soil with about 5 cm maize stem was ploughed to a depth of approximately 20 cm in April each year.Each cropping cycle, therefore, consisted of periods of maize (from May to September) and fallow (from October to April) of the following year.The precipitation and air temperature data were acquired from the meteorological station of the Shenyang Agro-Ecological Station. The precipitation during the 2012/2013, 2013/2014, 2014/2015, and 2015/2016 periods were 911.9 mm, 621.7 mm, 485.7 mm, and 585.3 mm, respectively (Fig. 1). 72.3%, 75.5%, 66.5%, and 73.0% of these annual precipitations occurred during maize-growing period, respectively. The mean annual air temperatures in these years were 7.7 °C (− 21.2 to 27.5 °C), 8.1 °C (− 22.7 to 28.3 °C), 9.5 °C (− 21.7 to 28.2 °C) and 9.3 °C (− 17.1 to 27.0 °C), respectively. The soil temperature at a depth of 5 cm varied between − 14 and 35 °C during the four-year period (Fig. 2b). The change trend of soil surface temperature was the same as that of soil temperature at 5 cm depth (Fig. 2a). The mean soil WFPS (0–15 cm) varied between 15 and 73% (Fig. 2c).Figure 1Precipitation and daily mean air temperature during four annual cycles from May 2012 to April 2016 in the experimental field.Full size imageFigure 2Seasonal variations in soil temperature (at soil surface and 5 cm soil depth) and WFPS% at 0–15 cm depth from May 2012 to April 2016.Full size imageGas sampling and analysisThe gas was sampled between 3rd May 2012 and 14th April 2016 using a static closed chamber system as described by Dong et al.16. Briefly, a stainless-steel chamber base (56 cm length × 28 cm width) was inserted into the soil of each plot to a depth of approximately 10 cm, with its long edge perpendicular to the rows of maize. The top chamber (56 cm length × 28 cm width × 20 cm height) was also made of stainless steel. Gas samples were obtained using a syringe 0, 20, and 40 min after the chambers had been closed between 9:00 am and 11:00 am on each sampling day. Gas samples were collected every 2‒6 days and every 7‒15 days during the growing seasons and non-growing seasons, respectively. The first gas sampling time was on day 1, day 3, day 1, and day 3 after maize planting each year. The N2O concentrations in gas samples were quantified using a gas chromatograph (Agilent 7890A, Shanghai, China) with an electron capture detector.Soil analysisThe soil temperature and volumetric water content (SVWC) were measured at depth of 0–15 cm using a bent stem thermometer and a time-domain reflectometry (Zhongtian Devices Co. Ltd, China), respectively. SVWC was converted to soil water-filled pore space (WFPS) using the following equation:$${text{WFPS}} = {text{SVWC}}/(1{-}{text{BD}}/{text{particle}},{text{density}}),$$
(1)
where BD is soil bulk density (g cm−3). Particle density was assumed to be 2.65 g cm−3.Soil samples from the 0–20 cm layer were collected in each plot in April 2012 (before sowing) and October 2015 (maize harvest) using a 5 cm diameter stainless steel soil sampler. The five soil samples collected from different locations in each plot were mixed thoroughly. Visible roots were removed by hand and the samples were air-dried and sieved using a 0.15 mm sieve. SOC was then quantified using an elemental analyzer (Vario EL III, Elementar, Germany). Soil available NH4+–N and NO3−–N were extracted with 2 M KCl and measured colorimetrically using a continuous flow injection analyzer (Futura, Alliance, France)17. Soil Olsen-P was extracted with NaHCO3 and colorimetrically measured using a spectrophotometer (Lambda 2, PerkinElmer, USA). Soil available K was extracted by 1 M CH3COONH4 and analyzed with a flame photometer (FP640, Jingmi, China). Soil pH was determined with deionized water (1:2.5) and analyzed using a pH meter (PHS-3C, LeiCi, China) with a glass electrode.DNA extraction and real-time quantitative PCRThe soil samples for measuring the abundance of nitrification and denitrification functional genes were collected on May 20, 2015. Soil DNA was extracted with the soil DNA extracted kits (EZNA soil DNA Kit; Omega Bio-Tek Inc., U.S.A.). The copy numbers of nitrification and denitrification functional genes were determined by q-PCR with the Roche LightCyler® 96 (Roche, Switzerland). Additional details about the primers and amplification procedure can be found in Dong et al.16.Data analysisThe N2O flux (μg N2O–N m−2 h−1) is calculated based on the increase of N2O concentration per unit chamber area for a specific time interval18 as follows:$${text{F}} = 273/left( {273 + {text{T}}} right) times {text{M}}/22.4 times {text{H}} times {text{dc}}/{text{dt}} times 1000$$
(2)
where F (μg N2O–N m−2 h−1) is the N2O flux, T (◦C) is the air temperature in the chamber, M (g N2O–N mol−1) is the molecular weight of N2O–N, 22.4 (L mol−1) is the molecular volume of the gas at 101.325 kPa and 273 K, H (m) is the chamber height, dc/dt (ppb h−1) is the rate of change in the N2O concentration in the chamber.Cumulative N2O emissions were calculated as follows:$${text{Cumulative}},{text{emission}} = mathop sum limits_{{{text{i}} = 1}}^{{text{n}}} frac{{({text{F}}_{{text{i}}} + {text{F}}_{i + 1} )}}{2} times ({text{t}}_{{{text{i}} + 1}} – {text{t}}_{{text{i}}} ) times 24$$
(3)
where F is the N2O emission flux (μg N2O–N m−2 h−1), i is the ith measurement, (ti+1 − ti) is the number of days between two adjacent measurements, and n is the total number of the measurements. Annual N2O emissions were calculated between the fertilization dates of each successive year.The SOC stock (Mg ha−1) in the topsoil was calculated as:$${text{C}}_{{{text{stock}}}} = {text{SOC}} times {text{BD}} times {text{D}} times 10,$$
(4)
where BD is soil bulk density (g cm−3), D is the depth of the topsoil (0.2 m).The topsoil SOC sequestration rate (SOCSR) (Mg ha−1 yr−1) was estimated using the following equation:$${text{SOCSR}} = left( {{text{C}}_{{{text{stock2015}}}} – {text{C}}_{{{text{stock2012}}}} } right) times {text{t}}^{ – 1} ,$$
(5)
where Cstock2015 and Cstock2012 are the SOC stocks in 2015 and 2012, respectively, and t is the duration of the experiment (years).Statistical analyses were performed using SPSS 13.0 (SPSS, Chicago, USA). The differences in cumulative N2O emissions and maize yields within a year, and other factors among treatments were assessed using one-way Analysis of Variance (ANOVA) with least significant difference post-hoc tests and a 95% confidence limit. The effects of different treatments, years, and their interactions on N2O emission, maize yield and aboveground biomass were examined using one-way repeated measures ANOVA. Pearson correlation analysis was used to analyze the relationships between cumulative N2O emissions and precipitation (N = 12 (three data each year, four years)), as well as N2O flux and soil available nitrogen content.
Statements of research involving plantsIt is stated that the current research on the plants comply with the relevant institutional, national, and international guidelines and legislation. It is also stated that the appropriate permissions have been taken wherever necessary, for collection of plant or seed specimens. It is also stated that the authors comply with the ‘IUCN Policy Statement on Research Involving Species at Risk of Extinction’ and the ‘Convention on the Trade in Endangered Species of Wild Fauna and Flora’. More

Today, awareness of our perilous position has grown immensely, even if our ability to do something about it has not. Analyses suggest that human activities have already pushed planetary processes past stable boundaries through destruction of biodiversity, ocean acidification, and land-use change associated with agriculture, among other effects (see Steffen, W. et al., Science 347, 1259855; 2015). Over the past few decades, estimates find that human resource extraction has reduced the total outstanding capital of the world’s base of natural resources by some 40%. What is apparently our most pressing challenge — planetary warming — is just one of many challenges linked to our inability to limit the scale of our human activities and impacts.
Your institute does not have access to this article More

Tomasello, M. Why we cooperate (MIT Press, 2009).Book 

Google Scholar 
Turchin, P. The puzzle of human ultrasociality: How did large-scale complex societies evolve? In Cultural Evolution, Strüngmann Forum Report Vol. 12 (eds Richerson, P. J. & Christiansen, M. H.) 61–73 (MIT Press, 2013).
Google Scholar 
Kramer, K. L. How there got to be so many of us: The evolutionary story of population growth and a life history of cooperation. J. Anthropol. Res. 75, 472–497 (2019).Article 

Google Scholar 
Wrangham, R. W. The Goodness Paradox: The Strange Relationship Between Virtue and Violence in Human Evolution (Alfred A. Knopf, 2019).
Google Scholar 
Fruth, B. & Hohmann, G. Food sharing across borders. Hum. Nat. 29, 91–103 (2018).PubMed 
PubMed Central 
Article 

Google Scholar 
Garfield, Z. H., Hubbard, R. L. & Hagen, E. H. Evolutionary models of leadership. Hum. Nat. 30, 23–58 (2019).PubMed 
Article 

Google Scholar 
Rodrigues, J. & Hewig, J. Let´ s call it altruism! A psychological perspective and hierarchical framework of altruism and prosocial behavior. Preprint at https://psyarxiv.com/pj7eu/ (2021).Davies, A. Food sharing. In Routledge Handbook of Sustainable and Regenerative Food Systems (eds Duncan, J. et al.) 204–217 (Routledge, 2020).Chapter 

Google Scholar 
Ember, C. R., Skoggard, I., Ringen, E. J. & Farrer, M. Our better nature: Does resource stress predict beyond-household sharing?. Evol. Hum. Behav. 39, 380–391 (2018).Article 

Google Scholar 
Crittenden, A. N. & Schnorr, S. L. Current views on hunter-gatherer nutrition and the evolution of the human diet. Am. J. Phys. Anthropol. 162, 84–109 (2017).PubMed 
Article 

Google Scholar 
Ferguson, M. et al. Traditional food availability and consumption in remote Aboriginal communities in the Northern Territory, Australia. Aust. NZ. J. Publ. Heal. 41, 294–298 (2017).Article 

Google Scholar 
Poulain, J. P. The Sociology of Food: Eating and the Place of Food in Society (Bloomsbury Publishing, 2017).
Google Scholar 
Ready, E. & Power, E. A. Why wage earners hunt: food sharing, social structure, and influence in an Arctic mixed economy. Curr. Anthropol. 59, 74–97 (2018).Article 

Google Scholar 
Gould, R. A. To have and have not: The ecology of sharing among hunter-gatherers. In Resource Managers: North American and Australian Hunter-Gatherers (eds Williams, N. M. & Hunn, E. S.) 69–91 (Routledge, 2019).Chapter 

Google Scholar 
Allen-Arave, W., Gurven, M. & Hill, K. Reciprocal altruism, rather than kin selection, maintains nepotistic food transfers on an Ache reservation. Evol. Hum. Behav. 29, 305–318 (2008).Article 

Google Scholar 
Crittenden, A. N. & Zes, D. A. Food sharing among hadza hunter-gatherer children. PLoS One 10, e0131996. https://doi.org/10.1371/journal.pone.0131996 (2015).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Rochat, P. et al. Fairness in distributive justice by 3-and 5-year-olds across seven cultures. J. Cross. Cult. Psychol. 40, 416–442 (2009).Article 

Google Scholar 
Cashdan, E. A. Coping with risk: Reciprocity among the Basarwa of Northern Botswana. Man 20, 454 (1985).Article 

Google Scholar 
Fehr, E., Glätzle-Rützler, D. & Sutter, M. The development of egalitarianism, altruism, spite and parochialism in childhood and adolescence. Eur. Econ. Rev. 64, 369–383 (2013).Article 

Google Scholar 
Almås, I., Cappelen, A. W., Sørensen, E. Ø. & Tungodden, B. Fairness and the development of inequality acceptance. Science 328, 1176–1178 (2010).ADS 
PubMed 
Article 
CAS 

Google Scholar 
Malti, T. et al. “Who is worthy of my generosity?” Recipient characteristics and the development of children’s sharing. Int. J. Behav. Dev. 40, 31–40 (2016).Article 

Google Scholar 
Olson, K. R. & Spelke, E. S. Foundations of cooperation in young children. Cognition 108, 222–231 (2008).PubMed 
PubMed Central 
Article 

Google Scholar 
Renno, M. P. & Shutts, K. Children’s social category-based giving and its correlates: expectations and preferences. Dev. Psychol. 51, 533 (2015).PubMed 
Article 

Google Scholar 
Samek, A. et al. The development of social comparisons and sharing behavior across 12 countries. J. Exp. Child Psychol. 192, 104778. https://doi.org/10.1016/j.jecp.2019.104778 (2020).Article 
PubMed 

Google Scholar 
Henrich, J., Heine, S. J. & Norenzayan, A. Most people are not WEIRD. Nature 466, 29–29 (2010).ADS 
CAS 
PubMed 
Article 

Google Scholar 
House, B. R. et al. Ontogeny of prosocial behavior across diverse societies. P. Natl. Acad. Sci. USA 110, 14586–14591 (2013).ADS 
CAS 
Article 

Google Scholar 
Schäfer, M., Haun, D. B. & Tomasello, M. Fair is not fair everywhere. Psychol. Sci. 26, 1252–1260 (2015).PubMed 
Article 

Google Scholar 
Callaghan, T. & Corbit, J. Early prosocial development across cultures. Curr. Opin. Psychol. 20, 102–106 (2018).PubMed 
Article 

Google Scholar 
Rodriguez, L. M., Martí-Vilar, M., Esparza Reig, J. & Mesurado, B. Empathy as a predictor of prosocial behavior and the perceived seriousness of delinquent acts: A cross-cultural comparison of Argentina and Spain. Ethics Behav. 31, 91–101 (2021).Article 

Google Scholar 
Fehr, E., Bernhard, H. & Rockenbach, B. Egalitarianism in young children. Nature 454, 1079–1083 (2008).ADS 
CAS 
PubMed 
Article 

Google Scholar 
Henrich, J. & Muthukrishna, M. The origins and psychology of human cooperation. Ann. Rev. Psychol. 72, 207–240 (2021).Article 

Google Scholar 
Thomas, M. G. et al. Kinship underlies costly cooperation in Mosuo villages. Roy. Soc. Open Sci. 5(2), 171535. https://doi.org/10.1098/rsos.171535 (2018).ADS 
Article 

Google Scholar 
O’Gorman, R., Sheldon, K. M. & Wilson, D. S. For the good of the group? Exploring group-level evolutionary adaptations using multilevel selection theory. Group. Dyn. Theor. Res. 12, 17 (2008).Article 

Google Scholar 
Boyd, R. & Richerson, P. J. Culture and the evolution of human cooperation. Philos. Trans. R. Soc. B 364, 3281–3288 (2009).Article 

Google Scholar 
Handley, C. & Mathew, S. Human large-scale cooperation as a product of competition between cultural groups. Nat. Commun. 11, 1–9 (2020).Article 
CAS 

Google Scholar 
Gintis, H., van Schaik, C. & Boehm, C. Zoon politikon: The evolutionary origins of human socio-political systems. Behav. Process. 161, 17–30 (2019).Article 

Google Scholar 
Markovits, H., Benenson, J. F. & Kramer, D. L. Children and adolescents’ internal models of food-sharing behavior include complex evaluations of contextual factors. Child Dev. 74, 1697–1708 (2003).PubMed 
Article 

Google Scholar 
Kaplan, H., Gurven, M., Hill, K. & Hurtado, A. M. The natural history of human food sharing and cooperation: a review and a new multi-individual approach to the negotiation of norms. Moral Sentim. Mater. Interests Found. Coop. Econ. Life 6, 75–113 (2005).
Google Scholar 
Crittenden, A. N. To share or not to share? Social processes of learning to share food among Hadza hunter-gatherer children. In Social Learning and Innovation in Contemporary Hunter-Gatherers (eds Hewlett, B. S. & Terashima, H.) 61–70 (Springer, 2016).Chapter 

Google Scholar 
Barragan, R. C., Brooks, R. & Meltzoff, A. N. Altruistic food sharing behavior by human infants after a hunger manipulation. Sci. Rep. 10, 1–9 (2020).Article 
CAS 

Google Scholar 
Singh, M., Wrangham, R. & Glowacki, L. Self-interest and the design of rules. Hum. Nat. 28, 457–480 (2017).PubMed 
Article 

Google Scholar 
Richerson, P. J., Gavrilets, S. & de Waal, F. B. Modern theories of human evolution foreshadowed by Darwin’s Descent of Man. Science 372, eaba3776. https://doi.org/10.1126/science.aba3776 (2021).CAS 
Article 
PubMed 

Google Scholar 
Jordan, F. M. et al. Cultural evolution of the structure of human groups. In Cultural Evolution: Society, Technology, Language, and Religion (eds Richerson, P. J. & Christiansen, M. H.) 87–116 (MIT Press, 2013).Chapter 

Google Scholar 
Henrich, J. & Broesch, J. On the nature of cultural transmission networks: Evidence from Fijian villages for adaptive learning biases. Philos. T. Roy. Soc. B 366, 1139–1148 (2011).Article 

Google Scholar 
Hawley, P. H. The ontogenesis of social dominance: A strategy-based evolutionary perspective. Dev. Rev. 19, 97–132 (1999).Article 

Google Scholar 
Hawley, P. H., Little, T. D. & Card, N. A. The allure of a mean friend: Relationship quality and processes of aggressive adolescents with prosocial skills. Int. J. Behav. Dev. 31, 170–180 (2007).Article 

Google Scholar 
Marlowe, F. The Hadza: Hunter-Gatherers of Tanzania Vol. 3 (University of California Press, 2010).
Google Scholar 
Jones, N. B. Demography and Evolutionary Ecology of Hadza Hunter-Gatherers Vol. 71 (Cambridge University Press, 2016).
Google Scholar 
Butovskaya, M. L. Aggression and conflict resolution among the nomadic Hadza of Tanzania as compared with their pastoralist neighbors. In War, Peace, and Human Nature: the Convergence of Evolutionary and Cultural Views (ed. Fry, D. P.) 278–296 (Oxford University Press, 2013).Chapter 

Google Scholar 
Apicella, C. L., Marlowe, F. W., Fowler, J. H. & Christakis, N. A. Social networks and cooperation in hunter-gatherers. Nature 481, 497–501 (2012).ADS 
CAS 
PubMed 
PubMed Central 
Article 

Google Scholar 
Sands, B., Maddieson, J. & Ladefoged, P. The phonetic structures of Hadza. Stud. Afr. Linguist. 25, 171–204 (1996).Article 

Google Scholar 
Butovskaya, M. et al. Approach to resource management and physical strength predict differences in helping: evidence from two small-scale societies. Front. Psychol. 11, 373 (2020).PubMed 
PubMed Central 
Article 

Google Scholar 
Mous, M. A Grammar of Iraqw (University of Leiden, 1992).
Google Scholar 
Rekdal, O. B. The invention by tradition: Creativity and change among the Iraqw of northern Tanzania. PhD thesis, Department of Social Anthropology, University of Bergen, Bergen (1999).Snyder, K. A. The Iraqw of Tanzania: Negotiating Rural Development (Routledge, 2018).Book 

Google Scholar 
Butovskaya, M., Burkova, V. & Mabulla, A. Sex differences in 2D: 4D ratio, aggression and conflict resolution in African children and adolescents: a cross-cultural study. J. Aggress. Confl. Peace Res. 2, 17–31 (2010).Article 

Google Scholar 
Butovskaya, M. L., Burkova, V. N. & Karelin, D. V. The Wameru of Tanzania: Historical origin and their role in the process of National Integration. Soc. Evol. Hist. 15, 141–163 (2016).
Google Scholar 
Lerner, G. The Creation of Patriarchy Vol. 1 (Oxford University Press, 1986).
Google Scholar 
Maruo, S. Differentiation of subsistence farming patterns among the Haya banana growers in northwestern Tanzania. Afr. Study Monog. 23, 147–175 (2002).
Google Scholar 
Ishengoma, J. M. African oral traditions: Riddles among the Haya of Northwestern Tanzania. Int. Rev. Educ. 51, 139–153 (2005).Article 

Google Scholar 
Stevens, L. Religious change in a Haya village, Tanzania. J. Relig. Afr. 21, 2–25 (1991).Article 

Google Scholar 
Kradin, N. N. The transformation of pastoralism in Buryatia: the Aginsky Steppe example. Inner Asia 6, 95–109 (2004).Article 

Google Scholar 
Rostovtseva, V. V., Weissing, F. J., Mezentseva, A. A. & Butovskaya, M. L. Sex differences in cooperativeness—an experiment with Buryats in Southern Siberia. PLoS One 15, e0239129. https://doi.org/10.1371/journal.pone.0239129 (2020).CAS 
Article 
PubMed 
PubMed Central 

Google Scholar 
Krader, L. Buryat religion and society. Southwest. J. Anthropol. 10, 322–351 (1954).Article 

Google Scholar 
Hooper, P. L. Quantitative description of the pastoral economy of Western Tuvan nomads. New Res. Tuva 4, 19–27 (2020).
Google Scholar 
Lindquist, G. Loyalty and command: Shamans, lamas, and spirits in a Siberian ritual. Soc. Anal. 52, 111–126 (2008).
Google Scholar 
Walters, P. Religion in Tuva: Restoration or innovation?. Relig. State Soc. 29, 23–38 (2001).Article 

Google Scholar 
Dyrtyk-ool, A. O., & Orgezhik, C. M. Kollektsiya vostochnih tuvintsev-olenevodov v natsionalnom muzee Respubliki Tyva: istoriya komplektovaniya i obshaya harakteristika [Collection of the Eastern Tuvans – deer herders in the National Museum of the Republic of Tuva: Background and general description of the acquisition]. Scientific notes of the museum-reserve “Tomskaya Pisanitsa”. 3, 4–9 (2016).Alexandrov, V. A., Vlasova, I. V. & Polischuk, N. S. The Russians (Nauka, 1997).
Google Scholar 
Fehr, E. & Schmidt, K. M. A theory of fairness, competition, and cooperation. Q. J. Econ. 114, 817–868 (1999).MATH 
Article 

Google Scholar 
Charness, G. & Rabin, M. Understanding social preferences with simple tests. J. Q. Econ. 117, 817–869 (2002).MATH 
Article 

Google Scholar  More

Phylogenetic relationships inside the family Palaemonidae remain unresolved, despite being frequently discussed in recent publications9,10. Nevertheless, the last published study5 presented the main lineages of the family as well supported. Among those, the studied Pon-I group of predominantly free-living taxa is basal-positioned to the remaining genera of the former subfamily Pontoniinae, usually more specialised and associated with a wide range of hosts. The basal separation of the symbiotic genera led some authors to consider the assemblage, following Bruce22, to be a primitive group, or descendants of such7,23. Additionally, Gan et al.8 suggested that the taxa of the Pon-I group might be direct descendants of the ancestors of the former subfamily Pontoniinae, sharing the main plesiomorphies appearing frequently in former palaemonine taxa, e.g., the genera Brachycarpus, Leptocarpus, Macrobrachium, or Palaemon. The median process on the fourth thoracic sternite can be considered a plesiomorphic feature; indeed, it is a common symplesiomorphy of all Pon-I taxa, including Ischnopontonia and Anapontonia, for which the process was formerly reported as missing24 (its presence was confirmed in present examined specimens). In addition to that, the mandibular palp occurring in the genera Exoclimenella, Eupontonia, Palaemonella, and Vir25, or the presence of two arthrobranchs on the third maxilliped in Exoclimenella26, can also be considered plesiomorphic features.The Pon-I group’s internal relations have been unclear until now due to lower generic and species coverage in previous studies4,5,8. The present analysis based on a six-marker molecular dataset allows a deeper insight into the phylogenetic relationships of the study group involving all 11 currently recognised genera, and represented by 52 species, i.e. about 60% of the overall known species diversity of the group. The results provide a strong support for the monophyly and/or taxonomic validity of the current genera Exoclimenella, Anapontonia, Ischnopontonia, and suggest the monophyly of genera Harpilius and Philarius. Moreover, the results reveal non-monophyly of the most speciose genera Palaemonella and Cuapetes, as well as the species-poor Eupontonia. The genus Palaemonella was found to be paraphyletic owing to the nested species of the genera Eupontonia and Vir, which all share a common synapomorphy, the presence of the mandibular palp (mentioned above). Such conclusion was expressed also in the study of Chow et al.5.The present phylogenetic analysis confirmed that the genus Cuapetes is not monophyletic, as found to a lesser extent, in a few previous molecular studies 4,5,23. In this study, the genus Cuapetes was recovered in four separate genetic lineages. The type species C. nilandensis is nested in the Clade 1 along with C. johnsoni and C. seychellensis. This phylogenetic finding is in line with the study of Marin and Sinelnikov27, who indicated morphological differences between two of the above-mentioned species and most of the remaining species of the genus (respective of the present Clade 5, also covering C. grandis, the type species of the ex-genus Kemponia), and questioned the validity of the two latter generic names. The further genetic lineage is shown by the position of C. americanus nested in the eastern Pacific—Atlantic branch of the genus Palaemonella (Clade 3). This result is also supported by recent phylogenetic studies suggesting the different systematic positions of this species4,5,10. Due to the lack of the mandibular palp, the species had been properly, but evidently incorrectly, assigned to the genus Cuapetes. The fourth genetic lineage is shown by a separate position of C. darwiniensis in the Clade 4 as the sister species of Madangella altirostris.The remaining majority of the Cuapetes species (Clade 5) are heterogeneous due to comprising also representatives of the genus Periclimenella. Ďuriš and Bruce26 hypothesised, based on morphological traits (mainly the unique shape of the first pereiopod chelae and the distinctly asymmetrical and specific second pereiopods), that the genera Exoclimenella and Periclimenella are closely related. Nevertheless, the present study revealed Periclimenella as a part of the genus Cuapetes. This result was previously supported in the molecular study by Horká et al.4 and weakly supported by Kou et al.23.Fossil records of palaemonid shrimps are rare due to their aquatic habit and poorly calcified exoskeletons. Only a few palaemonid representatives are known compared to many extant taxa; the oldest fossil records contain only genera from the previous subfamily Palaemoninae from the Lower Cretaceous (middle Albian, 100 Myr)28. For this reason, we used the known mutation rate of mitochondrial gene (16S rRNA) for dating rather than fossil records.The present inferred phylogeny and ancestral analysis indicate multiple formations of primary symbioses within the clades dominated by free-living relatives, as shown by previous molecular analyses4,5. Our results revealed eight independent lineages within the Pon-I group that evolved from free-living ancestors (Fig. 3). Free-living palaemonids (Exoclimenella, Palaemonella, Cuapetes; Fig. 2) are characterised by an elongate body shape with a dentate rostrum, slender, long, a/symmetrical chelipeds and slender ambulatory pereiopods with simple dactyli. Their carapace might bear the full complement of teeth (i.e., supraorbital, antennal, hepatic, epigastric)25. Primary symbiotic forms do not fundamentally differ morphologically from free-living ancestors. Their adaptations to the host affiliation have mainly manifested by changes in body shape, colouration, and the reduction of carapace ornamentation. Their hosts belong to different invertebrate phyla, including Cnidaria (mainly Scleractinia and Antipatharia22) and Echinodermata (Crinoidea29) in ectosymbiotic forms, but also to spoon worms (Echiura), burrowing Crustacea (alpheid shrimps), and/or gobiid fishes15, in inquilinistic forms.While scleractinian corals were hypothesised as the primary hosts of palaemonid shrimp commensalism7, our results revealed the antipatharian association as possibly the earlier one among the Pon-I shrimps. That association was established via a single speciation act at approximately 43 Myr (Eocene), specifically with the ancestor of the recent Cuapetes nilandensis (Clade 1). Except a small body size, this species does not show specific morphological adaptations to antipatharian association. The possibly oldest lineage associated with the scleractinian corals forms a common multigeneric composition of Anapontonia, Ischnopontonia, Harpilius and Philarius (Clade 4), which was established at approximately 38.2 Myr (Eocene). The genera share some homoplasic adaptations with ectosymbioses, such as strongly hooked dactyli of the ambulatory pereiopods adapted to climbing on coral colonies. An extremely compressed body and similar tail fan structure of the genera Ischnopontonia (Fig. 1H) and Anapontonia (Fig. 1D) are adaptations to life in narrow spaces amongst corallites of the oculinid coral Galaxea24,30; the intercorallite channels might be temporarily fully covered by tentacles of exposed polyps. This lifestyle was thus termed ‘semi-endosymbiosis’ by Horká et al.4, as potential evolutionary precursors of the true endosymbioses. In contrast, the genera Philarius and Harpilius have depressed bodies and associate exclusively as regular ectosymbionts with scleractinian corals, mainly of the genera Acropora and Pocillopora22.A further multispecies symbiotic lineage is represented by the genus Vir (Clade 3), whose origin is dated to approximately 21.1 Myr (Miocene). All species of this genus live in associations mainly with the acroporid, pocilloporid and euphylliid genera of scleractinian corals31,32. The adaptation to their symbiotic lifestyle is expressed in the loss of the hepatic tooth, partial or full reduction of ambulatory propodal spines, and cryptic colouration, including transparency of the body and appendages31,33 (Fig. 1J). Subsequent scleractinian-associated lineages are represented by separate species that appeared in the Miocene (21.9–10.1 Myr), namely: Eupontonia oahu, Cuapetes amymone, and C. kororensis, which live in association with Pocillopora, Acropora, and Heliofungia, and show only minor adaptations to their symbiotic habits, e.g. loss of the hepatic tooth, dense distal setae on the walking propodi, or extremely slender chelae and a specific cryptic colouration, respectively22,34,35.A single crinoid-associated species, Palaemonella pottsi (Clade 3), represents the only case of the switch from a free-living lifestyle to the association with echinoderms in the present study group; it originated at approximately 10.4 Myr (Miocene). Retaining the body shape typical for Palaemonella12, the species also does not show any noticeable morphological adaptation to such a host; its affiliation with the symbiotic life is, however, clearly observed in the deep-red to black cryptic colouration36.In Palaemonella aliska (Fig. 1E) and Eupontonia nudirostris (Clade 3), a pair of sister-positioned species in the present analyses (Figs. 2, 3), the ability to co-habit with burrowing animals (e.g., alpheids, gobiid fish, or echiurids) had developed. Their type of symbiosis, inquilinism, formed at approximately 14.8 Myr (Miocene). The reduction of the rostrum length, depressed body, stout main chelae in both, and full lack of the epigastric and hepatic teeth in the latter species15,25, were evidently due to that mode of life. Inquilinism is best known in the family Alpheidae, in which multiple genera associate with a variety of burrowing animals37. In the family Palaemonidae, inquilinism developed only in the Pon-I group, including Palaemonella shirakawai (not analysed here)14.As evident from the present and previously published reports4,5,7,8,10, the life history of the Pon-I group was largely shaped by coevolution with coral reefs. The coral reefs were deeply impacted by the K–T mass extinction at the end of the Cretaceous, which was one of the most destructive events in the Phanerozoic38. However, coral reefs recovered and became increasingly abundant in the Eocene39. This also matches the time of either the origin of host associations, or a wider species radiation of the Pon-I group. The first fossil records of the main coral hosts of the present shrimps are dated after the K-T extinction during the Paleogene (e.g., Euphyllia 66.0–61.6 Myr, Acropora 59.2–56.0 Myr, Galaxea and Pocillopora 56–33.9 Myr40).The biogeographic history suggested by S-DIVA analysis points to some dispersal and vicariant events shaping the current pattern of the Pon-I group’s distribution. This reconstruction (Fig. 4) estimates the present-day IWP region within the former Paleo-Tethys Ocean as the most likely ancestral area of the present study group, which originated ~ 91.6 Myr (Late to Early Cretaceous). The present shrimp group had radiated across the entire IWP region and subsequently expanded into the Atlantic Ocean. We assume that the spread of the group took place in the following sequence of events: (1) dispersal of Palaemonella spp. from the IWP into the eastern Pacific in the Paleocene (∼ 55.2 Myr; P. asymmetrica and P. holmesi); (2) dispersal into the western Atlantic (2 spp., complex of “Cuapetes” americanus) via the eastern Pacific and vicariance event separating the IWP at Eocene (∼ 46.2 Myr). It was the time after the formation of the Eastern Pacific Barrier (EPB), which was considered the largest extension of the open ocean (ca. 5000 km), that separated the IWP area from the eastern Pacific17; (3) the another vicariance event, separating the western Atlantic populations from those of the eastern Pacific in the Oligocene (∼ 30.9 Myr), i.e., before the closure of the Isthmus of Panama, followed by a dispersion of P. atlantica into the eastern Atlantic in the Miocene (∼ 21.6 Myr). The exact time of the formation of the Isthmus of Panama, which separated the Atlantic from the eastern Pacific and remained isolated from the central Pacific by the EPB, still remains questionable. Bacon et al.18 assume that the initial land bridge formed at approximately 23 Myr, and the final closure of the Isthmus of Panama formed between 10 and 6 Myr. Montes et al.19 presupposed the earlier formation of the barrier at ∼ 14 Myr, whereas O’Dea et al.20 concluded that the potential gene flow continued between the Pacific and Atlantic subpopulations of marine organisms until at least ∼ 2.8 Myr.The eastern Pacific Cuapetes canariensis closely related to IWP Cuapetes spp., has been recently described by Fransen et al.41, from the Canary Islands. This could indicate alternative dispersal pathways into the Atlantic, as suggested by recent studies17,42. The Tethys seaway allowed natural dispersion between the Atlantic and Indian Oceans across the region of the Mediterranean Sea. The closure of this interoceanic seaway at approximately 14 Myr (18–12 Myr) was caused by intense tectonic activity in the Near East17. Since the closure of that seaway, remaining possible dispersal to the Atlantic has been limited to the warm-water corridor around the southern tip of Africa, however curtailed by the cold Benguela Current upwelling from the Late Pliocene43. More

This study proposed a protocol to evaluate the behavior of geese reared outdoors in agroforestry systems. A data collection form (i.e., BOF) was developed and validated both in relation to its reliability and its validity. In this context, moreover, ABMs useful for a welfare assessment protocol could be defined, and changes in the behavior of geese due to daily time and environmental context could be identified.Behavioral observations, based on the capture of the major changes in an animal’s body language17, are used daily in the assessment of animal health and welfare. Body language is a type of dynamic expression of the interactions among conspecifics or between animals and their environment. Behavioral changes can happen quickly or as subtle shifts not easily detectable18. Indeed, especially in the case of direct observation in the field, it becomes difficult to identify each behavioral variation. Furthermore, the on-farm use of the BOF proposed in the present study involved focal subgroup sampling, as ten geese were simultaneously observed, which may increase the difficulties. Indirect observation by videos, which allow the review of a certain action several times and the focal-animal approach, is a useful tool to partially overcome these issues and thus improve the accuracy of observation. The validation process of the BOF adopted in this study, therefore, included the definition of both its interobserver reliability and correlation with indirect observations.In this study, the direct observations in the field were performed by both an expert (i.e., main observer) and an inexperienced trained observer. As expected, the main observer was able to detect a higher frequency of behaviors, especially the rarer ones. For example, the inexperienced observer did not report any examples of allo-grooming, squawking, wagging tail, stretching, or panting behavior. However, the two observers showed excellent interobserver reliability (ICC  > 0.75). Major agreements were found for walking, roosting, and foraging. Accordingly, several studies have shown that observers with little experience can also provide a valuable contribution in observational research19,20. Overall, these results support the reliability of the BOF even if the observer’s experience helps him or her to better grasp rarer behaviors, as these behaviors could play an important role as welfare indicators.In the last two decades, important technological developments have occurred in the livestock sector. The use of sensors, cameras, and other devices can generate objective information about individual behavior, thereby allowing its evaluation in large observation areas and for large groups of animals and resulting in the better detection of natural animal behavior. Thus, in our study, the data collected by a video recording system (Noldus XT) were used as a gold standard measure to define the criterion validity of the BOF. Our results indicated excellent agreement between direct and indirect observations, supporting the BOF criterion validity. A poor correlation was only found for 2 variables (i.e., squawking and wagging tail), which were more difficult to collect by direct observation. The use of the BOF involved the simultaneous observation of 10 animals, but the geese had a synchronized behavior and moved in groups within the grazing area. This greatly facilitated focal subgroup sampling, allowed all animals to always be under observation, and could explain the high correlation between the two observation methods. However, the comparison between the observations collected in the field by the main observer and those recorded using the computerized system confirmed the greater accuracy of the latter. The analysis of the video in continuous with the use of some tools, such as the zoom or slow-motion functions, and the focal-animal sampling provided an easier identification of some behaviors and, in general, greater accuracy. Due to its nonintrusive approach, video recording has become a common practice for behavior assessment21, but it can be expensive and time-consuming. On the other hand, direct observations made by the BOF were valid and less expensive, suggesting that it could be a feasible tool with which to evaluate the welfare principle of Appropriate behavior. As recommended for welfare assessment protocols22, the BOF ethogram included indicators of both positive and negative states; however, it would be necessary to integrate it with behavioral tests and other ABMs evaluating the human-animal relationship.As mentioned above, there is no standardized geese behavior ethogram. Thus, to verify the content validity of the BOF, its behavior variables were analyzed through a PCA. The 4 extracted PCs could represent the broad behavioral dimensions of geese. In particular, the geese’s activity reported in PC1 was characterized by locomotor, foraging, and exploratory behaviors, with opposite signs with respect to roosting. The positive correlation between explorative and grazing activities and their negative correlation with static behaviors has been widely demonstrated in chickens. Chicken genotypes characterized by low exploratory aptitude exhibited low kinetic behaviors but a high frequency of roost and rest behaviors23. Göransson et al.24 showed that 50% of the observed birds exhibited sitting behavior, whereas less than 10% performed foraging activity.PC2 included all the variables that characterized the geese’s social aspects, including both positive and negative interactions. Usually, greylag geese live in a large flock because the offspring remain with their parents for an entire year. Such groups are characterized by complex relationships based on social interactions25. The formation of a group is characterized by agonistic behaviors such as fighting, pecking, and threatening, as well as submissive behaviors such as avoiding contact, crouching, and escaping26 to establish a hierarchical order. After this phase, a tolerance status develops, and birds maintain their social interactions through the use of body postures and vocalizations. Accordingly, the variables reported in PC2 were related not only to aggressive behaviors but also to geese’s vocalization and posture, which probably helped to maintain flock stability. Therefore, a higher PC2 score could indicate the need to establish and maintain a hierarchical order within the group, resulting in high social interactions.PC3 reported comfort and body care behaviors. The opportunity to spend a lot of time on body care, which should also include access to water for bathing, is of paramount importance with regard to fulfilling the biological requirements of geese27. Thus, a higher loading of this PC means that animals showed a good degree of both welfare and adaptability. In our study, a high frequency of self-cleaning and wing flapping behaviors was recorded, and the geese often took advantage of the water tub. In contrast, a very low frequency of aggression behaviors was observed, suggesting that the groups of geese were quite stable and that the animals felt safe in the environment in which they were rear. These findings confirm that agroforestry has a favorable impact on bird welfare by allowing the display of the full range of behavior, improving the animals’ comfort28.PC4 was mainly represented by the neck forward behavior. This position only occasionally represents an attack behavior and is not utilized during the establishment of hierarchical order but when it is necessary to maintain and reinforce the order inside the group. Furthermore, a goose that assumes this posture often does so while continuing another activity29. The neck forward behavior was positively associated with the stretching behavior. Stretching is usually categorized as a comfort behavior for broilers30, but it could also be used when the animal needs to relax stress-related tension in their muscles31,32 or as an adaptive strategy for dealing with unknown contexts33. Neck forward and stretching were eventually considered social avoidance behaviors, although they could be ambivalent and thus require further study, case-by-case assessment, and perhaps a better description in the ethogram.Finally, some interesting results emerged regarding the comparison of geese’s behavior during the morning and afternoon and between the two different agroforestry systems. In particular, geese showed a higher frequency of active behaviors such as walking, foraging, drinking, neck forward, and feeding during the morning compared to the afternoon. All of these behaviors suggest that geese concentrate their grazing and exploration activities during the morning. When and where to move is crucial for the food search and to avoid both predators and adverse climate conditions34. Cartoni Mancinelli et al.35 included exploratory attitude, walking, and eating grass activities in a multifactorial score as important parameters to consider to evaluate the adaptability of different organically reared chicken genotypes. Thus, exploratory and kinetic behaviors are fundamental, especially in animals reared outdoors. Moreover, the positive correlation between walking and grazing behaviors is widely known36,37. In contrast, during the afternoon, geese showed higher frequencies of static behaviors such as resting, roosting, and self-grooming, suggesting that geese are more dedicated to comfort and body care activities during this time. These trials were performed in the hottest season; thus, the geese’s behavioral differences during the day could also depend on the fact that animals preferred to carry out active behaviors during the cooler hours (morning), while in the hottest hours (afternoon), they engaged in static activities. Active behaviors cause an increase in metabolism and body temperature38, whereas static behavior, such as roosting, is considered adaptative behavior to promote heat dissipation31,39.This could also explain why higher frequencies of walking and foraging and lower frequencies of static behaviors were found in the orchard system than in the vineyard system. Studies carried out on chickens have reported that, among different pasture enrichments, the presence of trees promotes walking animal activity compared with crop inclusion40,41. The cover provided by trees made the animals feel protected from predators and provided shade during the hottest part of the day40, thereby stimulating the animals to explore all the available space in the pen. Accordingly, geese reared in the apple orchard ingested more grass than those reared in a vineyard36. However, there were no differences between the two systems for social behaviors. Moreover, the highest frequency of roosting and self-cleanliness behaviors was recorded in the vineyard, suggesting that this space offered a comfortable environment and that both systems seem respectful of the biological needs and welfare of the geese.The behavioral assessment protocol proposed in this study involving the BOF ethogram was feasible, low-cost, fast, and responsive both over time and between housing systems. It could thus be used for the assessment of Appropriate behavior in a welfare assessment protocol for geese reared in outdoor or free-range systems, although it lacks indicators of the human-animal relationship, such as avoidance distance or handling tests; such a scoring system should be developed. Regarding the specific behaviors in the two agroforestry systems, it should also be noted that they are difficult to generalize, as the characteristics of the plants, the environment, and management could have influenced these traits. Specifically, the behaviors could have been affected by the temperatures; therefore, further trials at different altitudes, seasons (i.e., autumn and winter), and climate are necessary for external validation. More

In line with the results of present study, we suggest that the exploitation of forage resources by agro-pastoralists occurs in a non-random manner. The use of forage resources is guided by a series of functional characters related to palatability and nutritional value, which determine preferential use due to the better quality of resource. At the same time, we understand that forage uses are complex and multifactorial in nature, and regulated in a substantial way by seasonality and ecological factors (Fig. 5), such as the availability of plant resources and local diversity.Figure 5Diagrammatic representation for the effects of vegetation cover and seasonality on forage resource selection in Dry Forests. Image created with Microsoft Office 2019 PowerPoint (www.office.com).Full size imageThe differences of plant species cited between areas reveal the positive effect of vegetation cover on the use and knowledge of plants by agro-pastoralists. Our findings reveal that the greater number of plant species mentioned by agro-pastoralists in Area II is directly associated with greater availability of resources in this area, as long as we consider vegetation cover as availability of resources, which allows different species to be used throughout the year. On the other hand, in regions with low vegetation cover (Area I), the low availability of resources limits the use and knowledge of plants by residents, which can lead to greater pressure on a small set of available species. Such findings reinforce the importance of vegetation cover for ecosystem provision of goods and services to human populations that depend directly or indirectly on these services.The most represented families found in the present study have also been reported in several other ethnobotanical studies6,16,17,29, with emphasis on Fabaceae and Poaceae, which are recognized for their high forage potential, which derives, above all, from high palatability and nutritional value30. Simultaneously, citations mostly for native species reflect the importance and potential of Caatinga resources as important components of the ruminant diet11, both for the woody and herbaceous strata, corroborating the estimate in the literature that 70% of vegetation has potential use as forage31.The characteristic seasonality of vegetation, on the other hand, represents a limiting factor for forage productivity, culminating in high fluctuations in quality and availability, as well as changes in the dominance of different strata and composition of forage species throughout the seasons11,32. The seasonal distribution of species explains the similarity of seasons between areas, with a higher similarity percentage for the dry seasons, since there is less availability of resources to be exploited compared to the rainy season. In this context, the potentially used species are commonly accessible woody species in both areas. However, during the rainy season, the high availability of herbaceous plants regulates different uses (Fig. 4), but even so, they also exhibit relatively similar patterns, mainly due to the woody component that denotes the common demand by ruminants at the beginning of this season.The effect of climatic variables on vegetation use patterns was documented by16,17, both of which showed greater richness in the use of herbaceous forage during the rainy season, a finding that reflects the seasonal distribution—restriction to that season—and decrease in the qualitative character of annual species33. At the same time, it also reflects the greater number of unique species for the rainy season. However, when compared to woody strata, significant differences in terms of richness are not found because although the diversity of herbaceous species in the Caatinga is greater24, it is much less known than that of the tree-shrub stratum11.Agro-pastoralists even characterize animal preferences for herbaceous stratum, but as its diversity is immense and ephemeral, they claim to have limited ability to identify the species. The high abundance of resources in the rainy season also reduces the concern with forage use, which implies less attention to the species that are consumed. In contrast, woody species, due to multiple uses and greater availability over time, tend to be better known10,34, with a different effect in the dry season making the optimal foraging pattern in this period inherent to the knowledge of agro-pastoralists35.In addition, according to the ecological appearance hypothesis, there is a general tendency for less apparent species to be neglected by populations36. Some studies have corroborated the hypothesis within the context of forage use, with woody species being cited more and having more uses6,15. In addition, people tend to focus on resources whose supply is given continuously10, which may explain why woody species are well represented in both seasons.Security in the provisioning of ecosystem services is an essential component for local populations, and thus woody species are highly valued because they reflect predictability of use15,35. This can be a particularly influential criterion because perennial or late leaf deciduous species, such as Cynophalla flexuosa and Myracrodruon urundeva, had significant amounts of citations and perceptions employing high valuation, as represented by some statements by some interviewees: “É um refrigero na seca” (it is savage in the dry season), “É uma ração boa na seca” (it is a good food in the dry season).In turn, differences in richness of the species cited by the two areas corroborate our first hypothesis that populations inserted in environments with greater vegetation cover tend to cite more species. In line with these findings, considerable floristic dissimilarity was also found between the two areas, given the exclusivity of species. Such dissimilarity may suggest particularities in the vegetation attributes of each area, such as greater floristic diversity7,37,38.Since anthropic processes are irregularly distributed in space, variation in the provisioning of ecosystem services by vegetation also occurs, and influences different collection profiles39. On the other hand, areas with greater species richness have been shown to have greater use patterns6,7. The larger number of species cited as woody and native for Area II is, therefore, associated with greater general richness, as well as herbaceous species present in the rainy season. In contrast, common species are reflected in trends of similar foraging patterns, as well as the presence of common species between areas38. In addition to different levels of disturbance, differences in floristic composition between areas may also be due to edaphic variation40.Our second hypothesis was refuted because the difference in the richness of exotic species between the areas. Plausible explanations for this finding are that, in general, exotic herbaceous species are commonly used for forage in the semi-arid region of Brazil41. Herbaceous species comprise the primary component of the ruminant diet. However, in the midst of their occurrence restricted to the short rainy period, exotic species, mainly of Fabaceae and Poaceae, have been introduced to increase the forage availability, which currently represents an important attribute of forage resources in the Caatinga41,42,43. At the same time, and to also increase the availability of forage resources, the cultivation of species by agro-pastoralists may be common in their properties44, mainly exotics, such as Prosopis juliflora, that have high adaptive potential and governmental incentives45.Regarding use patterns, according to the data presented here it is possible to state that agro-pastoralists ’ experiences with herding activities provide an accumulation of a vast knowledge about forage resources15. This knowledge allows forage resources to be characterized by their potential according to a variety of criteria associated with seasonal variation and qualitative attributes, as commonly found by other studies14,15,16,17,37. Such criteria are often revealed by qualitative approaches that define the valuation perception of resources. Thus, nutritional value and palatability can be implicitly associated with the definitions of “É uma ração boa” (it is a good food), “o bicho gosta muito” (the animals like it very much) and “Rico em proteínas” (rich in protein).It should be added that the establishment of intrinsic relationships with resources allows a particular understanding at a high level of detail15,35, such as changes in palatability throughout development with descriptions including chemical17 and structural changes. Studies confirm that some Caatinga species vary in their chemical composition during leaf maturation, which influences nutritional quality17,46.In addition to revealing the domain of information, this body of knowledge allows maximizing forage use based on nutritional properties weighted by availability14,37. Nunes37 confirmed that the forage species selected by informants and the criteria they adopted coincided with nutritional values measured by the literature, and that, as also found in the present study, younger plants were recognized as highly appreciated by animals. This appreciation is due to the greater palatability of plant organs at this stage47. This is a matter of concern for the sustainability of the Caatinga, since direct or indirect grazing has compromised the regeneration process12 since younger individuals are clearly more sensitive to damage48.Also, considering the potential of Caatinga, we suggest that investment through government actions encourage the cultivation of native species to ensure the production of forage and, consequently, guarantee the sustainability of livestock activity and the ecosystem in question. More